Nash Q-Network for Multi-agent Cybersecurity Simulation

Cite

@InProceedings{10.1007/978-3-032-08067-7_3,
author=”Xie, Qintong
and Koh, Edward
and Cadet, Xavier
and Chin, Peter”,
editor=”Baras, John S.
and Papavassiliou, Symeon
and Tsiropoulou, Eirini Eleni
and Sayin, Muhammed O.”,
title=”Nash Q-Network for Multi-agent Cybersecurity Simulation”,
booktitle=”Game Theory and AI for Security”,
year=”2026″,
publisher=”Springer Nature Switzerland”,
address=”Cham”,
pages=”43–60″,
abstract=”Cybersecurity defense involves interactions between adversarial parties (namely defenders and hackers), making multi-agent reinforcement learning (MARL) an ideal approach for modeling and learning strategies for these scenarios. This paper addresses the challenge of simultaneous multi-agent training in complex environments and introduces a Nash Q-Network that enables learning in a partial observation environment. Facilitates learning in partially observed settings. We demonstrate the successful implementation of this algorithm in a notable complex cyber defense simulation treated as a two-player zero-sum Markov game setting. We propose the Nash Q-Network, which aims to learn Nash-optimal strategies that translate to robust defenses in cybersecurity settings. Our approach incorporates aspects of proximal policy optimization (PPO), deep Q-network (DQN), and the Nash-Q algorithm, addressing common challenges like non-stationarity and instability in multi-agent learning. The training process employs distributed data collection and carefully designed neural architectures for both agents and critics.”,
isbn=”978-3-032-08067-7″
}

Explore Reinforced: Equilibrium Approximation with Reinforcement Learning

Mateusz Nowak, Qintong Xie, Emma Graham, Ryan Yu, Michelle Yilin Feng, Roy Leibovitz, Xavier Cadet, Sang (Peter) Chin

Cite

@InProceedings{10.1007/978-3-032-08064-6_3,
author=”Nowak, Mateusz
and Xie, Qintong
and Graham, Emma
and Yu, Ryan
and Feng, Michelle Yilin
and Leibovitz, Roy
and Cadet, Xavier
and Chin, Peter”,
editor=”Baras, John S.
and Papavassiliou, Symeon
and Tsiropoulou, Eirini Eleni
and Sayin, Muhammed O.”,
title=”Explore Reinforced: Equilibrium Approximation with Reinforcement Learning”,
booktitle=”Game Theory and AI for Security”,
year=”2026″,
publisher=”Springer Nature Switzerland”,
address=”Cham”,
pages=”42–60″,
abstract=”Current approximate Coarse Correlated Equilibria (CCE) algorithms struggle with equilibrium approximation for games in large stochastic environments. While these game-theoretic methods are theoretically guaranteed to converge to a strong solution concept, reinforcement learning (RL) algorithms have shown increasing capability in such environments but lack the equilibrium guarantees provided by game-theoretic approaches. In this paper, we introduce Exp3-IXRL – an equilibrium approximator that utilizes RL, specifically leveraging the agent’s action selection, to update equilibrium approximations while preserving the integrity of both learning processes. We therefore extend the Exp3 algorithms beyond the stateless, non-stochastic settings. Empirically, we demonstrate improved performance in classic non-stochastic multi-armed bandit settings, capability in stochastic multi-armed bandits, and strong results in a complex and adversarial cybersecurity network environment.”,
isbn=”978-3-032-08064-6″
}

PoolFlip: A Multi-agent Reinforcement Learning Security Environment for Cyber Defense

Xavier Cadet, Simona Boboila, Sie Hendrata Dharmawan, Alina Oprea, Sang (Peter) Chin

Cite

@InProceedings{10.1007/978-3-032-08064-6_9,
author=”Cadet, Xavier
and Boboila, Simona
and Dharmawan, Sie Hendrata
and Oprea, Alina
and Chin, Peter”,
editor=”Baras, John S.
and Papavassiliou, Symeon
and Tsiropoulou, Eirini Eleni
and Sayin, Muhammed O.”,
title=”PoolFlip: A Multi-agent Reinforcement Learning Security Environment for Cyber Defense”,
booktitle=”Game Theory and AI for Security”,
year=”2026″,
publisher=”Springer Nature Switzerland”,
address=”Cham”,
pages=”172–192″,
abstract=”Cyber defense requires automating defensive decision-making under stealthy, deceptive, and continuously evolving adversarial strategies. The FlipIt game provides a foundational framework for modeling interactions between a defender and an advanced adversary that compromises a system without being immediately detected. In FlipIt, the attacker and defender compete to control a shared resource by performing a Flip action and paying a cost. However, the existing FlipIt frameworks rely on a small number of heuristics or specialized learning techniques, which can lead to brittleness and the inability to adapt to new attacks. To address these limitations, we introduce PoolFlip, a multi-agent gym environment that extends the FlipIt game to allow efficient learning for attackers and defenders. Furthermore, we propose Flip-PSRO, a multi-agent reinforcement learning (MARL) approach that leverages population-based training to train defender agents equipped to generalize against a range of unknown, potentially adaptive opponents. Our empirical results suggest that Flip-PSRO defenders are {\$}{\$}2{\backslash}times {\$}{\$}2{\texttimes}more effective than baselines to generalize to a heuristic attack not exposed in training. In addition, our newly designed ownership-based utility functions ensure that Flip-PSRO defenders maintain a high level of control while optimizing performance.”,
isbn=”978-3-032-08064-6″
}

Tree Search for Simultaneous Move Games via Equilibrium Approximation

Ryan Yu, Alex Olshevsky, Sang (Peter) Chin

Cite

@InProceedings{10.1007/978-3-032-08064-6_1,
author=”Yu, Ryan
and Olshevsky, Alex
and Chin, Peter”,
editor=”Baras, John S.
and Papavassiliou, Symeon
and Tsiropoulou, Eirini Eleni
and Sayin, Muhammed O.”,
title=”Tree Search for Simultaneous Move Games via Equilibrium Approximation”,
booktitle=”Game Theory and AI for Security”,
year=”2026″,
publisher=”Springer Nature Switzerland”,
address=”Cham”,
pages=”3–22″,
abstract=”Neural network supported tree-search has shown strong results in a variety of perfect information multi-agent tasks. However, the performance of these methods on imperfect information games has generally been below competing approaches. Here we study the class of simultaneous-move games, which are a subclass of imperfect information games which are most similar to perfect information games: both agents know the game state with the exception of the opponent’s move, which is revealed only after each agent makes its own move. Simultaneous move games include popular benchmarks such as Google Research Football and Starcraft Multi Agent Challenge. Our goal in this paper is to take tree search algorithms trained through self-play and adapt them to simultaneous move games without significant loss of performance. While naive ways to do this fail, we are able to achieve this by deriving a practical method that attempts to approximate a coarse correlated equilibrium as a subroutine within a tree search. Our algorithm, Neural Network-Coarse Correlated Equilibrium (NN-CCE), works on cooperative, competitive, and mixed tasks and our results are better than the current best MARL algorithms on a wide range of accepted baselines.”,
isbn=”978-3-032-08064-6″
}

Strategic Cyber Defense via Reinforcement Learning-Guided Combinatorial Auctions

Mai Pham, Vikrant Vaze, Sang (Peter) Chin

Cite

@INPROCEEDINGS{11196565,
author={Pham, Mai and Vaze, Vikrant and Chin, Peter},
booktitle={2025 IEEE High Performance Extreme Computing Conference (HPEC)},
title={Strategic Cyber Defense via Reinforcement Learning-Guided Combinatorial Auctions},
year={2025},
volume={},
number={},
pages={1-7},
abstract={Cyber defense operations increasingly require long-term strategic planning under uncertainty and resource constraints. We propose a new use of combinatorial auctions for allocating defensive action bundles in a realistic cyber environment, using host-specific valuations derived from reinforcement learning (RL) Q-values. These Q-values encode long-term expected utility, allowing upstream planning. We train CAFormer, a differentiable Transformer-based auction mechanism, to produce allocations that are approximately incentive-compatible under misreporting. Rather than benchmarking against existing agents, we explore the qualitative and strategic properties of the learned mechanisms. Compared to oracle and heuristic allocations, our method achieves competitive revenue while offering robustness to misreporting. In addition, we find that allocation patterns correlate with adversarial and defensive activity, suggesting implicit alignment with operational priorities. Our results demonstrate the viability of auction-based planning in cyber defense and highlight the interpretability benefits of RL-derived value structures.},
keywords={Training;Uncertainty;Reinforcement learning;Strategic planning;Transformers;Robustness;Resource management;Cost accounting;Optimization;Resilience;Cyber defense;strategic planning;mechanism design;differentiable optimization},
doi={10.1109/HPEC67600.2025.11196565},
ISSN={2643-1971},
month={Sep.},}

Degree matrix comparison for graph alignment

Ashley Wang, Sang (Peter) Chin

Cite

@INPROCEEDINGS{11196278,
author={Wang, Ashley and Chin, Peter},
booktitle={2025 IEEE High Performance Extreme Computing Conference (HPEC)},
title={Degree Matrix Comparison for Graph Alignment},
year={2025},
volume={},
number={},
pages={1-7},
abstract={The graph alignment problem, which considers the optimal node correspondence across networks, has recently gained significant attention due to its wide applications. There are various graph alignment methods, but we focus on the unsupervised geometric alignment algorithms. We propose Degree Matrix Comparison (DMC), a degree-based method that has shown to be effective for heterogeneous networks. Through extensive experiments and mathematical proofs, we demonstrate the potential of this method. Remarkably, DMC achieves up to 99% correct node alignment for 90%-overlap networks and 100% accuracy for isomorphic graphs. Additionally, we propose a reduced Greedy DMC with lower time complexity and Weighted DMC that has demonstrated potential for aligning weighted graphs. Positive results from applying Greedy DMC and the Weighted DMC furthermore speaks to the validity and potential of the DMC. The sequence of DMC methods could significantly impact graph alignment, offering reliable solutions for the task.},
keywords={Accuracy;Heterogeneous networks;Reliability;Time complexity;Graph Alignment;Unsupervised Method;Heterogeneous Networks},
doi={10.1109/HPEC67600.2025.11196278},
ISSN={2643-1971},
month={Sep.},}

Quantitative Resilience Modeling for Autonomous Cyber Defense

Xavier Cadet, Simona Boboila, Edward Koh, Sang (Peter) Chin, Alina Oprea

Cite

@article{cadet2025quantitative,
title={Quantitative Resilience Modeling for Autonomous Cyber Defense},
author={Cadet, Xavier and Boboila, Simona and Koh, Edward and Chin, Peter and Oprea, Alina},
journal={Reinforcement Learning Journal},
volume={6},
pages={894–908},
year={2025}
}

ZipNN: Lossless Compression for AI Models

Moshik Hershcovitch, Andrew Wood, Leshem Choshen, Guy Girmonsky, Roy Leibovitz, Or Ozeri, Ilias Ennmouri, Michal Malka, Sang (Peter) Chin, Swaminathan Sundararaman, Danny Harnik

Cite

@INPROCEEDINGS{11120577,
author={Hershcovitch, Moshik and Wood, Andrew and Choshen, Leshem and Girmonsky, Guy and Leibovitz, Roy and Ozeri, Or and Ennmouri, Ilias and Malka, Michal and Chin, Peter and Sundararaman, Swaminathan and Harnik, Danny},
booktitle={2025 IEEE 18th International Conference on Cloud Computing (CLOUD)},
title={ZipNN: Lossless Compression for AI Models},
year={2025},
volume={},
number={},
pages={186-198},
abstract={With the growth of model sizes and the scale of their deployment, their sheer size burdens the infrastructure requiring more network and more storage to accommodate these. While there is a vast model compression literature deleting parts of the model weights for faster inference, we investigate a more traditional type of compression – one that represents the model in a compact form and is coupled with a decompression algorithm that returns it to its original form and size – namely lossless compression. We present ZipNN, a lossless compression tailored to neural networks. Somewhat surprisingly, we show that specific lossless compression can gain significant network and storage reduction on popular models, often saving 33% and at times reducing over 50% of the model size. We investigate the source of model compressibility and introduce specialized compression variants tailored for models that further increase the effectiveness of compression. On popular models (e.g. Llama 3) ZipNN shows space savings that are over 17% better than vanilla compression while also improving compression and decompression speeds by 62%. Using multiple workers and threads, ZipNN can achieve decompression speeds of up to 80GB/s and compression speed of up to 13GB/s. We estimate that these methods could save over an ExaByte per year of network traffic downloaded from a large model hub like Hugging Face.},
keywords={Training;Cloud computing;Computational modeling;Redundancy;Neural networks;Telecommunication traffic;Logic gates;Artificial intelligence;Model compression;Faces;compression;lossless compression;models;AI;language models},
doi={10.1109/CLOUD67622.2025.00028},
ISSN={2159-6190},
month={July},}

Machine Learning-Assisted Multi-Wavelength Perception Enabled by Ultrathin, Large-Area Printed Indium Oxysulfide Phototransistors

Simon Agnew, Xavier Cadet, Sang (Peter) Chin, William Scheideler

Cite

@INPROCEEDINGS{11105739,
author={Agnew, Simon A. and Cadet, Xavier F. and Chin, Peter and Scheideler, William J.},
booktitle={2025 Device Research Conference (DRC)},
title={Machine Learning-Assisted Multi-Wavelength Perception Enabled by Ultrathin, Large-Area Printed Indium Oxysulfide Phototransistors},
year={2025},
volume={},
number={},
pages={1-2},
abstract={We present a method of fabricating uniform, large area indium oxysulfide ($\mathrm{InO}{\mathrm{x}} \mathrm{S}{\mathrm{y}}$) films using a vacuum-free continuous liquid metal printing method (CLMP) and sulfurization process for high-performance multi-wavelength photodetection. CLMP enables rapid printing of wide area ($\gt10 \mathrm{~cm}^{2} / \mathrm{s}$) metal oxide films of single nmscale thickness at process temperatures just above $150^{\circ} \mathrm{C}$, which can be partially converted to metal oxy-chalcogenide thin films at back-end-of line (BEOL) process temperatures. Phototransistors fabricated from 16 nm-thick $\mathrm{InO}{\mathrm{x}} \mathrm{S}{\mathrm{y}}$ achieved responsivities as high as $280 \mathrm{~A} / \mathrm{W}$ and respond to wavelengths as long as 630 nm, enabling both classification of multiple wavelengths and readout of intensity assisted by machine learning models.},
keywords={Printing;Liquids;Films;Machine learning;Indium;Phototransistors},
doi={10.1109/DRC66027.2025.11105739},
ISSN={2640-6853},
month={June},}

Implementation of a machine learning model and direct-to-patient outreach program for targeted screening for familial hypercholesterolemia

Kerrilynn Hennessey, Shoshana Bardach, Terry Sturke, Vikrant Vaze, Roshni Kalkur, Adam Prince, Hanyuan Shi, Marc Hofley, Sang (Peter) Chin, Rachel Forcino, Mary McGowan

Cite

@article{HENNESSEY20251029,
title = {Implementation of a machine learning model and direct-to-patient outreach program for targeted screening for familial hypercholesterolemia},
journal = {Journal of Clinical Lipidology},
volume = {19},
number = {4},
pages = {1029-1036},
year = {2025},
issn = {1933-2874},
doi = {https://doi.org/10.1016/j.jacl.2025.04.192},
url = {https://www.sciencedirect.com/science/article/pii/S1933287425002612},
author = {Kerrilynn C. Hennessey and Shoshana H. Bardach and Terry Sturke and Vikrant S. Vaze and Roshni S. Kalkur and Adam J. Prince and Hanyuan Shi and Marc A. Hofley and Peter Chin and Rachel Forcino and Mary P. McGowan},
keywords = {Machine learning, Familial hypercholesterolemia, Implementation science, Outreach, Artificial intelligence, Screening, Prevention},
abstract = {OBJECTIVES
Heterozygous familial hypercholesterolemia (FH) is underdiagnosed. This program evaluated the impact of implementing a machine learning model (MLM), expert chart review and clinical consultation in the diagnosis of FH.
METHODS
Flag, Identify, Network and Deliver FH (FIND-FH) was applied to 147,412 unique patient records in the Dartmouth Health system and identified 388 adult patients at risk for FH. Lipidologists and cardiologists performed chart reviews using FH clinical criteria. Patients were excluded from outreach if they had an established diagnosis of FH, a lipidologist, insufficient information to suspect FH, a low likelihood of FH, moved, died, or had alternative medical priorities at the time of review.
RESULTS
Among 388 flagged patients, median age was 50 years (IQR: 39-59 years), 43% were female, and 88% self-identified as white. After expert review, 208 (54%) patients were removed from outreach for meeting exclusion criteria. The majority of those excluded had a low likelihood of having FH (115/208, 55%). The median low-density lipoprotein cholesterol (LDL-C) in excluded patients was 134 mg/dL (IQR: 102-154 mg/dL) compared to 172 mg/dL (IQR: 132-216 mg/dL) in patients selected for outreach. A high-touch, direct-to-patient outreach process yielded 72 clinical visits (19%) and 58 new diagnoses of possible/probable/definite FH (15%).
CONCLUSION
The Find-FH MLM flagged 388 individuals as “at risk” for FH of whom 58 (15%) ultimately received a diagnosis of possible/probable/definite FH. While this represents a substantial improvement on 1:250 (0.4%) expected when screening the general population, it was labor intensive. For scalability, improved accuracy of the MLM and efficiency of chart review and outreach are needed.}
}

Can Language Models Take A Hint? Prompting for Controllable Contextualized Commonsense Inference

Publication:

Cite

@misc{colonhernandez2024languagemodelshintprompting,
title={Can Language Models Take A Hint? Prompting for Controllable Contextualized Commonsense Inference},
author={Pedro Colon-Hernandez and Nanxi Liu and Chelsea Joe and Peter Chin and Claire Yin and Henry Lieberman and Yida Xin and Cynthia Breazeal},
year={2024},
eprint={2410.02202},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2410.02202},
}

SocioDojo: Building Lifelong Analytical Agents with Real-world Text and Time Series

Cite

@inproceedings{
anonymous2024sociodojo,
title={SocioDojo: Building Lifelong Analytical Agents with Real-world Text and Time Series},
author={Anonymous},
booktitle={The Twelfth International Conference on Learning Representations},
year={2024},
url={https://openreview.net/forum?id=s9z0HzWJJp}
}

Bridging Neural and Symbolic Representations with Transitional Dictionary Learning

Cite

@inproceedings{
anonymous2024bridging,
title={Bridging Neural and Symbolic Representations with Transitional Dictionary Learning},
author={Anonymous},
booktitle={The Twelfth International Conference on Learning Representations},
year={2024},
url={https://openreview.net/forum?id=uqxBTcWRnj}
}

Detecting Continuous Gravitational Waves Using Generated Training Data

Cite

@INPROCEEDINGS{10445938,
author={Herrmann, Judith and Kunert, Raphael and Hachmon, Ron and Markus, Aviv and Gunby-Mann, Allison and Cohen, Sarel and Friedrich, Tobias and Chin, Peter},
booktitle={ICASSP 2024 – 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
title={Detecting Continuous Gravitational Waves Using Generated Training Data},
year={2024},
volume={},
number={},
pages={146-150},
keywords={Training;Gravitational waves;Training data;Signal processing;Task analysis;Speech processing;Synthetic data;Continuous Gravitational Waves Detection;Physics-Based Synthetic Data;SFTs;CNNs},
doi={10.1109/ICASSP48485.2024.10445938}}

Learned Approximate Distance Labels for Graphs

Cite

@InProceedings{10.1007/978-3-031-53468-3_29,
author=”Abioye, Ikeoluwa
and Gunby-Mann, Allison
and Wang, Xu
and Cohen, Sarel
and Chin, Peter”,
editor=”Cherifi, Hocine
and Rocha, Luis M.
and Cherifi, Chantal
and Donduran, Murat”,
title=”Learned Approximate Distance Labels for Graphs”,
booktitle=”Complex Networks {\&} Their Applications XII”,
year=”2024″,
publisher=”Springer Nature Switzerland”,
address=”Cham”,
pages=”339–350″,
abstract=”Distance computation is a fundamental problem in algorithmic graph theory with broad applications across various fields. Distance labeling is the method of assigning a label {\$}{\$}{\backslash}ell {\$}{\$}ℓto each node in a given graph G such that the distance between any pair of nodes u, v can be efficiently computed (or approximated) using only their labels {\$}{\$}{\backslash}ell (u){\$}{\$}ℓ(u)and {\$}{\$}{\backslash}ell (v){\$}{\$}ℓ(v). Minimizing the size of these labels is of crucial importance for performance. In this paper, we address this challenge by introducing a novel learning-based approach to distance labeling inspired by collaborative filtering. This approach achieves superior performance compared to the theoretical baseline on label size with a trade-off in distance approximation error on special graph classes such as cycles and trees. We also report promising experimental results on general graphs that obtain lower error than cycles and trees.”,
isbn=”978-3-031-53468-3″
}

Deep Distance Sensitivity Oracles

Cite

@misc{jeong2023deep,
title={Deep Distance Sensitivity Oracles},
author={Davin Jeong and Allison Gunby-Mann and Sarel Cohen and Maximilian Katzmann and Chau Pham and Arnav Bhakta and Tobias Friedrich and Sang Chin},
year={2023},
eprint={2211.02681},
archivePrefix={arXiv},
primaryClass={cs.LG}
}

ANEDA: Adaptable Node Embeddings for Shortest Path Distance Approximation

Cite

@INPROCEEDINGS{10363460,
author={Pacini, Frank and Gunby-Mann, Allison and Cohen, Sarel and Chin, Peter},
booktitle={2023 IEEE High Performance Extreme Computing Conference (HPEC)},
title={ANEDA: Adaptable Node Embeddings for Shortest Path Distance Approximation},
year={2023},
volume={},
number={},
pages={1-7},
keywords={Training;Deep learning;Roads;Heuristic algorithms;Measurement uncertainty;Area measurement;Routing;graph embedding;shortest path distance;neural networks},
doi={10.1109/HPEC58863.2023.10363460}}

Improved And Optimized Drug Repurposing For The SARS-CoV-2 Pandemic

Cite

@article{10.1371/journal.pone.0266572,
doi = {10.1371/journal.pone.0266572},
author = {Cohen, Sarel AND Hershcovitch, Moshik AND Taraz, Martin AND Kißig, Otto AND Issac, Davis AND Wood, Andrew AND Waddington, Daniel AND Chin, Peter AND Friedrich, Tobias},
journal = {PLOS ONE},
publisher = {Public Library of Science},
title = {Improved and optimized drug repurposing for the SARS-CoV-2 pandemic},
year = {2023},
month = {03},
volume = {18},
url = {https://doi.org/10.1371/journal.pone.0266572},
pages = {1-13},
abstract = {The active global SARS-CoV-2 pandemic caused more than 426 million cases and 5.8 million deaths worldwide. The development of completely new drugs for such a novel disease is a challenging, time intensive process. Despite researchers around the world working on this task, no effective treatments have been developed yet. This emphasizes the importance of drug repurposing, where treatments are found among existing drugs that are meant for different diseases. A common approach to this is based on knowledge graphs, that condense relationships between entities like drugs, diseases and genes. Graph neural networks (GNNs) can then be used for the task at hand by predicting links in such knowledge graphs. Expanding on state-of-the-art GNN research, Doshi et al. recently developed the Dr-COVID model. We further extend their work using additional output interpretation strategies. The best aggregation strategy derives a top-100 ranking of 8,070 candidate drugs, 32 of which are currently being tested in COVID-19-related clinical trials. Moreover, we present an alternative application for the model, the generation of additional candidates based on a given pre-selection of drug candidates using collaborative filtering. In addition, we improved the implementation of the Dr-COVID model by significantly shortening the inference and pre-processing time by exploiting data-parallelism. As drug repurposing is a task that requires high computation and memory resources, we further accelerate the post-processing phase using a new emerging hardware—we propose a new approach to leverage the use of high-capacity Non-Volatile Memory for aggregate drug ranking.},
number = {3},
}

A Study on Self-Supervised Object Detection Pretraining

Cite

@InProceedings{10.1007/978-3-031-25069-9_6,
author=”Dang, Trung
and Kornblith, Simon
and Nguyen, Huy Thong
and Chin, Peter
and Khademi, Maryam”,
editor=”Karlinsky, Leonid
and Michaeli, Tomer
and Nishino, Ko”,
title=”A Study on Self-Supervised Object Detection Pretraining”,
booktitle=”Computer Vision — ECCV 2022 Workshops”,
year=”2023″,
publisher=”Springer Nature Switzerland”,
address=”Cham”,
pages=”86–99″,
abstract=”In this work, we study different approaches to self-supervised pretraining of object detection models. We first design a general framework to learn a spatially consistent dense representation from an image, by randomly sampling and projecting boxes to each augmented view and maximizing the similarity between corresponding box features. We study existing design choices in the literature, such as box generation, feature extraction strategies, and using multiple views inspired by its success on instance-level image representation learning techniques [6, 7]. Our results suggest that the method is robust to different choices of hyperparameters, and using multiple views is not as effective as shown for instance-level image representation learning. We also design two auxiliary tasks to predict boxes in one view from their features in the other view, by (1) predicting boxes from the sampled set by using a contrastive loss, and (2) predicting box coordinates using a transformer, which potentially benefits downstream object detection tasks. We found that these tasks do not lead to better object detection performance when finetuning the pretrained model on labeled data.”,
isbn=”978-3-031-25069-9″
}

Semi-supervised adversarial text generation based on seq2seq models

Cite

@Inproceedings{Le2022,
author = {Hieu Le and Dieu-Thu Le and Verena Weber and Chris Church and Kay Rottmann and Melanie Bradford and Peter Chin},
title = {Semi-supervised adversarial text generation based on seq2seq models},
year = {2022},
url = {https://www.amazon.science/publications/semi-supervised-adversarial-text-generation-based-on-seq2seq-models},
booktitle = {EMNLP 2022},
}

Substitutional Neural Image Compression

Cite

@INPROCEEDINGS{wang-substitutional-2021,
author={Wang, Xiao and Ding, Ding and Jiang, Wei and Wang, Wei and Xu, Xiaozhong and Liu, Shan and Kulis, Brian and Chin, Peter},
booktitle={2022 Picture Coding Symposium (PCS)},
title={Substitutional Neural Image Compression},
year={2022},
volume={},
number={},
pages={97-101},
doi={10.1109/PCS56426.2022.10018005}}

cs-net: structural approach to time-series forecasting for high-dimensional feature space data with limited observations

Publication:

Cite

@misc{zong2022csnetstructuralapproachtimeseries,
title={cs-net: structural approach to time-series forecasting for high-dimensional feature space data with limited observations},
author={Weiyu Zong and Mingqian Feng and Griffin Heyrich and Peter Chin},
year={2022},
eprint={2212.02567},
archivePrefix={arXiv},
primaryClass={cs.LG},
url={https://arxiv.org/abs/2212.02567},
}

Towards Fast Crash-Consistent Cluster Checkpointing

Cite

@inproceedings{wood-towards-2022,
author = {Wood, Andrew and Hershcovitch, Moshik and Ennmouri, Ilias and Zong, Weiyu and Chennuri, Saurav and Cohen, Sarel and Sundararaman, Swaminathan and Waddington, Daniel and Chin, Peter},
booktitle = {2022 IEEE High Performance Extreme Computing Conference (HPEC)},
doi = {10.1109/HPEC55821.2022.9926330},
number = {},
pages = {1-8},
title = {Towards Fast Crash-Consistent Cluster Checkpointing},
volume = {},
year = {2022}
}

Biologically Plausible Complex-Valued Neural Networks and Model Optimization

Cite

@inproceedings{yu_biologically_2022,
abstract = {Artificial Neural Networks (ANNs) are thinly based on biological neural pathways. In an ANN, each node computes its activation by applying a non-linearity to a weighted sum of its inputs. While this formulation has been wildly successful for a variety of tasks, it is still a far cry from its biological counterpart, largely due to ANNs lack of phase information during computation. In this paper, we adapt ANNs to operate on complex values which naturally allows the inclusion of phase information during the forward pass. We demonstrate that our complex-valued architecture generally performs better compared to real-valued and other complex-valued networks in similar conditions. Additionally, we couple our model with a biologically inspired form of dimensionality reduction and present our findings on the MNIST and MusicNet data sets.},
address = {Cham},
author = {Yu, Ryan and Wood, Andrew and Cohen, Sarel and Hershcovitch, Moshick and Waddington, Daniel and Chin, Peter},
booktitle = {Artificial Intelligence Applications and Innovations},
editor = {Maglogiannis, Ilias and Iliadis, Lazaros and Macintyre, John and Cortez, Paulo},
isbn = {978-3-031-08333-4},
pages = {369–382},
publisher = {Springer International Publishing},
title = {Biologically Plausible Complex-Valued Neural Networks and Model Optimization},
year = {2022}
}

Training Robust Zero-Shot Voice Conversion Models with Self-Supervised Features

Cite

@inproceedings{dang_training_2022,
abstract = {Unsupervised Zero-Shot Voice Conversion (VC) aims to modify the speaker characteristic of an utterance to match an unseen target speaker without relying on parallel training data. Recently, self-supervised learning of speech representation has been shown to produce useful linguistic units without using transcripts, which can be directly passed to a VC model. In this paper, we showed that high-quality audio samples can be achieved by using a length resampling decoder, which enables the VC model to work in conjunction with different linguistic feature extractors and vocoders without requiring them to operate on the same sequence length. We showed that our method can outperform many baselines on the VCTK dataset. Without modifying the architecture, we further demonstrated that a) using pairs of different audio segments from the same speaker, b) adding a cycle consistency loss, and c) adding a speaker classification loss can help to learn a better speaker embedding. Our model trained on LibriTTS using these techniques achieves the best performance, producing audio samples transferred well to the target speaker’s voice, while preserving the linguistic content that is comparable with actual human utterances in terms of Character Error Rate.},
author = {Dang, Trung and Tran, Dung and Chin, Peter and Koishida, Kazuhito},
booktitle = {ICASSP 2022 – 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
doi = {10.1109/ICASSP43922.2022.9747743},
month = {May},
note = {ISSN: 2379-190X},
pages = {6557–6561},
title = {Training Robust Zero-Shot Voice Conversion Models with Self-Supervised Features},
year = {2022}
}

A Method to Reveal Speaker Identity in Distributed ASR Training, and How to Counter IT

Cite

@inproceedings{dang_method_2022,
abstract = {End-to-end Automatic Speech Recognition (ASR) models are commonly trained over spoken utterances using optimization methods like Stochastic Gradient Descent (SGD). In distributed settings like Federated Learning, model training requires transmission of gradients over a network. In this work, we design the first method for revealing the identity of the speaker of a training utterance with access only to a gradient. We propose Hessian-Free Gradients Matching, an input reconstruction technique that operates without second derivatives of the loss function (required in prior works), which can be expensive to compute. We show the effectiveness of our method using the DeepSpeech model architecture, demonstrating that it is possible to reveal the speaker’s identity with 34% top-1 accuracy (51% top-5 accuracy) on the LibriSpeech dataset. Further, we study the effect of Dropout on the success of our method. We show that a dropout rate of 0.2 can reduce the speaker identity accuracy to 0% top-1 (0.5% top-5).},
author = {Dang, Trung and Thakkar, Om and Ramaswamy, Swaroop and Mathews, Rajiv and Chin, Peter and Beaufays, Françoise},
booktitle = {ICASSP 2022 – 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
doi = {10.1109/ICASSP43922.2022.9746443},
month = {May},
note = {ISSN: 2379-190X},
pages = {4338–4342},
title = {A Method to Reveal Speaker Identity in Distributed ASR Training, and How to Counter IT},
year = {2022}
}

Referring Expressions with Rational Speech Act Framework: A Probabilistic Approach

Publication:

Cite

@misc{le_referring_2022,
abstract = {This paper focuses on a referring expression generation (REG) task in which the aim is to pick out an object in a complex visual scene. One common theoretical approach to this problem is to model the task as a two-agent cooperative scheme in which a “speaker” agent would generate the expression that best describes a targeted area and a “listener” agent would identify the target. Several recent REG systems have used deep learning approaches to represent the speaker/listener agents. The Rational Speech Act framework (RSA), a Bayesian approach to pragmatics that can predict human linguistic behavior quite accurately, has been shown to generate high quality and explainable expressions on toy datasets involving simple visual scenes. Its application to large scale problems, however, remains largely unexplored. This paper applies a combination of the probabilistic RSA framework and deep learning approaches to larger datasets involving complex visual scenes in a multi-step process with the aim of generating better-explained expressions. We carry out experiments on the RefCOCO and RefCOCO+ datasets and compare our approach with other end-to-end deep learning approaches as well as a variation of RSA to highlight our key contribution. Experimental results show that while achieving lower accuracy than SOTA deep learning methods, our approach outperforms similar RSA approach in human comprehension and has an advantage over end-to-end deep learning under limited data scenario. Lastly, we provide a detailed analysis on the expression generation process with concrete examples, thus providing a systematic view on error types and deficiencies in the generation process and identifying possible areas for future improvements.},
author = {Le, Hieu and Daryanto, Taufiq and Zhafransyah, Fabian and Wijaya, Derry and Coppock, Elizabeth and Chin, Sang},
keywords = {Computer Science – Computation and Language},
month = {May},
note = {arXiv:2205.07795 [cs]},
publisher = {arXiv},
shorttitle = {Referring Expressions with Rational Speech Act Framework},
title = {Referring Expressions with Rational Speech Act Framework: A Probabilistic Approach},
url = {http://arxiv.org/abs/2205.07795},
urldate = {2022-08-06},
year = {2022}
}

Collusion Detection in Team-Based Multiplayer Games

Publication:

Cite

@misc{greige_collusion_2022,
abstract = {In the context of competitive multiplayer games, collusion happens when two or more teams decide to collaborate towards a common goal, with the intention of gaining an unfair advantage from this cooperation. The task of identifying colluders from the player population is however infeasible to game designers due to the sheer size of the player population. In this paper, we propose a system that detects colluding behaviors in team-based multiplayer games and highlights the players that most likely exhibit colluding behaviors. The game designers then proceed to analyze a smaller subset of players and decide what action to take. For this reason, it is important and necessary to be extremely careful with false positives when automating the detection. The proposed method analyzes the players’ social relationships paired with their in-game behavioral patterns and, using tools from graph theory, infers a feature set that allows us to detect and measure the degree of collusion exhibited by each pair of players from opposing teams. We then automate the detection using Isolation Forest, an unsupervised learning technique specialized in highlighting outliers, and show the performance and efficiency of our approach on two real datasets, each with over 170,000 unique players and over 100,000 different matches.},
annote = {Comment: 14 pages, 4 figures},
author = {Greige, Laura and Silva, Fernando De Mesentier and Trotter, Meredith and Lawrence, Chris and Chin, Peter and Varadarajan, Dilip},
keywords = {Computer Science – Machine Learning, Computer Science – Computer Science and Game Theory},
month = {March},
note = {arXiv:2203.05121 [cs]},
publisher = {arXiv},
title = {Collusion Detection in Team-Based Multiplayer Games},
url = {http://arxiv.org/abs/2203.05121},
urldate = {2022-08-06},
year = {2022}
}

Non-Volatile Memory Accelerated Posterior Estimation

Publication:

Cite

@misc{wood_non-volatile_2022,
abstract = {Bayesian inference allows machine learning models to express uncertainty. Current machine learning models use only a single learnable parameter combination when making predictions, and as a result are highly overconfident when their predictions are wrong. To use more learnable parameter combinations efficiently, these samples must be drawn from the posterior distribution. Unfortunately computing the posterior directly is infeasible, so often researchers approximate it with a well known distribution such as a Gaussian. In this paper, we show that through the use of high-capacity persistent storage, models whose posterior distribution was too big to approximate are now feasible, leading to improved predictions in downstream tasks.},
author = {Wood, Andrew and Hershcovitch, Moshik and Waddington, Daniel and Cohen, Sarel and Chin, Peter},
keywords = {Computer Science – Machine Learning, Computer Science – Hardware Architecture},
month = {February},
note = {arXiv:2202.10522 [cs]},
publisher = {arXiv},
title = {Non-Volatile Memory Accelerated Posterior Estimation},
url = {http://arxiv.org/abs/2202.10522},
urldate = {2022-08-06},
year = {2022}
}

What Is Learned in Knowledge Graph Embeddings?

Cite

@inproceedings{douglas_what_2022,
abstract = {A knowledge graph (KG) is a data structure which represents entities and relations as the vertices and edges of a directed graph with edge types. KGs are an important primitive in modern machine learning and artificial intelligence. Embedding-based models, such as the seminal TransE [Bordes et al. 2013] and the recent PairRE [Chao et al. 202] are among the most popular and successful approaches for representing KGs and inferring missing edges (link completion). Their relative success is often credited in the literature to their ability to learn logical rules between the relations.},
address = {Cham},
author = {Douglas, Michael R. and Simkin, Michael and Ben-Eliezer, Omri and Wu, Tianqi and Chin, Peter and Dang, Trung V. and Wood, Andrew},
booktitle = {Complex Networks & Their Applications X},
editor = {Benito, Rosa Maria and Cherifi, Chantal and Cherifi, Hocine and Moro, Esteban and Rocha, Luis M. and Sales-Pardo, Marta},
isbn = {978-3-030-93413-2},
pages = {587–602},
publisher = {Springer International Publishing},
title = {What Is Learned in Knowledge Graph Embeddings?},
year = {2022}
}

Neural-Guided, Bidirectional Program Search for Abstraction and Reasoning

Cite

@inproceedings{alford_neural-guided_2022,
abstract = {One of the challenges facing artificial intelligence research today is designing systems capable of utilizing systematic reasoning to generalize to new tasks. The Abstraction and Reasoning Corpus (ARC) measures such a capability through a set of visual reasoning tasks. In this paper we report incremental progress on ARC and lay the foundations for two approaches to abstraction and reasoning not based in brute-force search. We first apply an existing program synthesis system called DreamCoder to create symbolic abstractions out of tasks solved so far, and show how it enables solving of progressively more challenging ARC tasks. Second, we design a reasoning algorithm motivated by the way humans approach ARC. Our algorithm constructs a search graph and reasons over this graph structure to discover task solutions. More specifically, we extend existing execution-guided program synthesis approaches with deductive reasoning based on function inverse semantics to enable a neural-guided bidirectional search algorithm. We demonstrate the effectiveness of the algorithm on three domains: ARC, 24-Game tasks, and a `double-and-add’ arithmetic puzzle.},
address = {Cham},
author = {Alford, Simon and Gandhi, Anshula and Rangamani, Akshay and Banburski, Andrzej and Wang, Tony and Dandekar, Sylee and Chin, John and Poggio, Tomaso and Chin, Peter},
booktitle = {Complex Networks & Their Applications X},
editor = {Benito, Rosa Maria and Cherifi, Chantal and Cherifi, Hocine and Moro, Esteban and Rocha, Luis M. and Sales-Pardo, Marta},
isbn = {978-3-030-93409-5},
pages = {657–668},
publisher = {Springer International Publishing},
title = {Neural-Guided, Bidirectional Program Search for Abstraction and Reasoning},
year = {2022}
}

Drug Repurposing Using Link Prediction on Knowledge Graphs with Applications to Non-volatile Memory

Cite

@inproceedings{cohen_drug_2022,
abstract = {The active global SARS-CoV-2 pandemic caused more than 167 million cases and 3.4 million deaths worldwide. The development of completely new drugs for such a novel disease is a challenging, time intensive process and despite researchers around the world working on this task, no effective treatments have been developed yet. This emphasizes the importance of drug repurposing, where treatments are found among existing drugs that are meant for different diseases. A common approach to this is based on knowledge graphs, that condense relationships between entities like drugs, diseases and genes. Graph neural networks (GNNs) can then be used for the task at hand by predicting links in such knowledge graphs. Expanding on state-of-the-art GNN research, Doshi et al. recently developed the Dr-COVID model. We further extend their work using additional output interpretation strategies. The best aggregation strategy derives a top-100 ranking of candidate drugs, 32 of which currently being in COVID-19-related clinical trials. Moreover, we present an alternative application for the model, the generation of additional candidates based on a given pre-selection of drug candidates using collaborative filtering. In addition, we improved the implementation of the Dr-COVID model by significantly shortening the inference and pre-processing time by exploiting data-parallelism. As drug repurposing is a task that requires high computation and memory resources, we further accelerate the post-processing phase using a new emerging hardware—we propose a new approach to leverage the use of high-capacity Non-Volatile Memory for aggregate drug ranking.},
address = {Cham},
author = {Cohen, Sarel and Hershcovitch, Moshik and Taraz, Martin and Kißig, Otto and Wood, Andrew and Waddington, Daniel and Chin, Peter and Friedrich, Tobias},
booktitle = {Complex Networks & Their Applications X},
editor = {Benito, Rosa Maria and Cherifi, Chantal and Cherifi, Hocine and Moro, Esteban and Rocha, Luis M. and Sales-Pardo, Marta},
isbn = {978-3-030-93413-2},
pages = {742–753},
publisher = {Springer International Publishing},
title = {Drug Repurposing Using Link Prediction on Knowledge Graphs with Applications to Non-volatile Memory},
year = {2022}
}

Deep Reinforcement Learning for FlipIt Security Game

Cite

@inproceedings{greige_deep_2022,
abstract = {Reinforcement learning has shown much success in games such as chess, backgammon and Go [21, 22, 24]. However, in most of these games, agents have full knowledge of the environment at all times. In this paper, we describe a deep learning model in which agents successfully adapt to different classes of opponents and learn the optimal counter-strategy using reinforcement learning in a game under partial observability. We apply our model to \$\$\backslashmathsf \FlipIt\$\$FlipIt[25], a two-player security game in which both players, the attacker and the defender, compete for ownership of a shared resource and only receive information on the current state of the game upon making a move. Our model is a deep neural network combined with Q-learning and is trained to maximize the defender’s time of ownership of the resource. Despite the noisy information, our model successfully learns a cost-effective counter-strategy outperforming its opponent’s strategies and shows the advantages of the use of deep reinforcement learning in game theoretic scenarios. We also extend \$\$\backslashmathsf \FlipIt\$\$FlipItto a larger action-spaced game with the introduction of a new lower-cost move and generalize the model to n-player \$\$\backslashmathsf \FlipIt\$\$FlipIt.},
address = {Cham},
author = {Greige, Laura and Chin, Peter},
booktitle = {Complex Networks & Their Applications X},
editor = {Benito, Rosa Maria and Cherifi, Chantal and Cherifi, Hocine and Moro, Esteban and Rocha, Luis M. and Sales-Pardo, Marta},
isbn = {978-3-030-93409-5},
pages = {831–843},
publisher = {Springer International Publishing},
title = {Deep Reinforcement Learning for FlipIt Security Game},
year = {2022}
}

Neural Network Optimization with Biologically Inspired Low-Dimensional Manifold Learning

Cite

@inproceedings{le_neural_2021,
abstract = {Neural Networks learn to recognize and leverage patterns in data. In most cases, while data is represented in a high-dimensional space, the patterns within the data exist along a manifold in a small subset of those dimensions. In this paper, we show that by using a biologically inspired algorithm called Geometric Multi-Resolution Analysis (GMRA), these low-dimensional manifolds can be computed and can be used to convert datasets into more useful forms for learning. We also show that, thanks to the lower-dimensional representation of the converted datasets, that smaller networks can achieve state-of-the-art performance while using significantly fewer parameters.},
author = {Le, Hieu and Wood, Andrew and Dandekar, Sylee and Chin, Peter},
booktitle = {2021 International Conference on Computational Science and Computational Intelligence (CSCI)},
doi = {10.1109/CSCI54926.2021.00006},
month = {December},
pages = {8–13},
title = {Neural Network Optimization with Biologically Inspired Low-Dimensional Manifold Learning},
year = {2021}
}

Corrupting Data to Remove Deceptive Perturbation: Using Preprocessing Method to Improve System Robustness

Cite

@inproceedings{le_corrupting_2021,
abstract = {Although deep neural networks have achieved great performance on classification tasks, recent studies showed that well trained networks can be fooled by adding subtle noises. This paper introduces a new approach to improve neural network robustness by applying the recovery process on top of the naturally trained classifier. In this approach, images will be intentionally corrupted by some significant operator and then be recovered before passing through the classifiers. SARGAN – an extension on Generative Adversarial Networks (GAN) is capable of denoising radar signals. This paper will show that SARGAN can also recover corrupted images by removing the adversarial effects. Our results show that this approach does improve the performance of naturally trained networks.},
author = {Le, Hieu and Walker, Hans and Tran, Dung and Chin, Peter},
booktitle = {2021 International Conference on Computational Science and Computational Intelligence (CSCI)},
doi = {10.1109/CSCI54926.2021.00308},
month = {December},
pages = {1594–1599},
title = {Corrupting Data to Remove Deceptive Perturbation: Using Preprocessing Method to Improve System Robustness},
year = {2021}
}

Mixed Spatio-Temporal Neural Networks on Real-time Prediction of Crimes

Cite

@inproceedings{zhou_mixed_2021,
abstract = {Forecasting the crime rate in real-time is always an important task to public safety. However, there are no known models that provide satisfactory approximation to this complex spatio-temporal problem until recently. The crime rate may be affected by various factors, such as local education, public events, weather, etc. Such factors make the prediction of crimes more complex and challenging than other problems that are less influenced by outer factors. In this paper, we propose a deep-learning-based approach, which combines various methods in neural networks to handle the spatial temporal prediction problem. Some optimization techniques, such as Bayesian optimization, are applied for finding the optimal hyper-parameters as well as dealing with noises in the dataset. The model is trained on a dataset about crime information in Los Angeles at a scale of hours in block-divided areas, released by the LA Police Department (LAPD). The results of experiments on this dataset demonstrates the proposed model’s ability in predicting potential crimes in real time.},
author = {Zhou, Xiao and Wang, Xiao and Brown, Gavin and Wang, Chengchen and Chin, Peter},
booktitle = {2021 20th IEEE International Conference on Machine Learning and Applications (ICMLA)},
doi = {10.1109/ICMLA52953.2021.00277},
month = {December},
pages = {1749–1754},
title = {Mixed Spatio-Temporal Neural Networks on Real-time Prediction of Crimes},
year = {2021}
}

How Dense Autoencoders can still Achieve the State-of-the-art in Time-Series Anomaly Detection

Cite

@inproceedings{jensen_how_2021,
abstract = {Time series data has become ubiquitous in the modern era of data collection. With the increase of these time series data streams, the demand for automatic time series anomaly detection has also increased. Automatic monitoring of data allows engineers to investigate only unusual behavior in their data streams. Despite this increase in demand for automatic time series anomaly detection, many popular methods fail to offer a general purpose solution. Some demand expensive labelling of anomalies, others require the data to follow certain assumed patterns, some have long and unstable training, and many suffer from high rates of false alarms. In this paper we demonstrate that simpler is often better, showing that a fully unsupervised multilayer perceptron autoencoder is able to outperform much more complicated models with only a few critical improvements. We offer improvements to help distinguish anomalous subsequences near to each other, and to distinguish anomalies even in the midst of changing distributions of data. We compare our model with state-of-the-art competitors on benchmark datasets sourced from NASA, Yahoo, and Numenta, achieving improvements beyond competitive models in all three datasets.},
author = {Jensen, Louis and Fosa, Jayme and Teitelbaum, Ben and Chin, Peter},
booktitle = {2021 20th IEEE International Conference on Machine Learning and Applications (ICMLA)},
doi = {10.1109/ICMLA52953.2021.00207},
month = {December},
pages = {1272–1277},
title = {How Dense Autoencoders can still Achieve the State-of-the-art in Time-Series Anomaly Detection},
year = {2021}
}

Revealing and Protecting Labels in Distributed Training

Cite

@inproceedings{dang_revealing_2021,
abstract = {Distributed learning paradigms such as federated learning often involve transmission of model updates, or gradients, over a network, thereby avoiding transmission of private data. However, it is possible for sensitive information about the training data to be revealed from such gradients. Prior works have demonstrated that labels can be revealed analytically from the last layer of certain models (e.g., ResNet), or they can be reconstructed jointly with model inputs by using Gradients Matching [Zhu et al.] with additional knowledge about the current state of the model. In this work, we propose a method to discover the set of labels of training samples from only the gradient of the last layer and the id to label mapping. Our method is applicable to a wide variety of model architectures across multiple domains. We demonstrate the effectiveness of our method for model training in two domains – image classification, and automatic speech recognition. Furthermore, we show that existing reconstruction techniques improve their efficacy when used in conjunction with our method. Conversely, we demonstrate that gradient quantization and sparsification can significantly reduce the success of the attack.},
author = {Dang, Trung and Thakkar, Om and Ramaswamy, Swaroop and Mathews, Rajiv and Chin, Peter and Beaufays, Françoise},
booktitle = {Advances in Neural Information Processing Systems},
editor = {Ranzato, M. and Beygelzimer, A. and Dauphin, Y. and Liang, P. S. and Vaughan, J. Wortman},
pages = {1727–1738},
publisher = {Curran Associates, Inc.},
title = {Revealing and Protecting Labels in Distributed Training},
url = {https://proceedings.neurips.cc/paper/2021/file/0d924f0e6b3fd0d91074c22727a53966-Paper.pdf},
volume = {34},
year = {2021}
}

TAK-ML: Applying Machine Learning at the Tactical Edge

Cite

@inproceedings{chin_tak-ml_2021,
abstract = {The “Every Soldier is a Sensor” (ES2) concept employs warfighters’ proximity to unfolding events in order to provide better situational awareness and decision-making capabilities. However, today’s ES2 practices put the burden of data collection on warfighters themselves, and the burden of interpretation (across potentially many inputs) on commanders. This leads to a situation where data collection is limited by the capacity of the warfighter (who is busy executing their core objectives), and data fusion, interpretation, and analysis are limited by the cognitive constraints of the human commanders and analysts interpreting the potentially massive amounts of data. The TAK-ML framework transitions these burdens to machines, allowing collection, fusion, and learning to operate at machine speed and scale. To accomplish this, TAK-ML takes recent advancements in mobile device capabilities and machine learning techniques and applies them to the Tactical Assault Kit (TAK) ecosystem, e.g., ATAK mobile devices and TAK servers, to facilitate the easy application of ML to real mission sets. This paper describes the TAK-ML framework which supports data collection, model building, and model execution/employment in tactical environments, as well as a set of initial applications of this framework. The framework and applications are described and evaluated, showing the capabilities available, the ease of use of the system, and initial insights into the efficacy of the resulting models and applications.},
author = {Chin, Peter and Do, Emily and Doucette, Cody and Kalashian, Brandon and Last, David and Lenz, Nathan and Lu, Edward and Minor, Devon and Noyes, Elias and Rock, Colleenn and Soule, Nathaniel and Walczak, Nicholas and Canestrare, Dave},
booktitle = {MILCOM 2021 – 2021 IEEE Military Communications Conference (MILCOM)},
doi = {10.1109/MILCOM52596.2021.9652909},
month = {November},
note = {ISSN: 2155-7586},
pages = {108–114},
title = {TAK-ML: Applying Machine Learning at the Tactical Edge},
year = {2021}
}

Intrinsic Examples: Robust Fingerprinting of Deep Neural Networks

Cite

@inproceedings{wang_intrinsic_2021,
abstract = {This paper proposes to use intrinsic examples as a DNN fingerprinting technique
for the functionality verification of DNN models implemented on edge devices. The
proposed intrinsic examples do not affect the normal DNN training and can enable the
black-box testing capability for DNN models packaged into edge device applications.
We provide three algorithms for deriving intrinsic examples of the pre-trained model
(the model before the DNN system design and implementation procedure) to retrieve
the knowledge learnt from the training dataset for the detection of adversarial third-party
attacks such as transfer learning and fault injection attack that may happen during the
system implementation procedure. Besides, they can accommodate the model transformations due to various DNN model compression methods used by the system designer.},
author = {Wang, Siyue and Zhao, Pu and Wang, Xiao and Chin, Sang Peter and Wahl, Thomas and Fei, Yunsi and Chen, Qi Alfred and Lin, Xue},
booktitle = {32nd British Machine Vision Conference 2021, BMVC 2021, Online, November 22-25, 2021},
pages = {46},
publisher = {BMVA Press},
title = {Intrinsic Examples: Robust Fingerprinting of Deep Neural Networks},
url = {https://www.bmvc2021-virtualconference.com/assets/papers/0625.pdf},
year = {2021}
}

PATCHCOMM: Using Commonsense Knowledge to Guide Syntactic Parsers

Cite

@inproceedings{xin_patchcomm_2021,
abstract = {Syntactic parsing technologies have become significantly more robust thanks to advancements in their underlying statistical and Deep Neural Network (DNN) techniques: most modern syntactic parsers can produce a syntactic parse tree for almost any sentence, including ones that may not be strictly grammatical. Despite improved robustness, such parsers still do not reflect the alternatives in parsing that are intrinsic in syntactic ambiguities. Two most notable such ambiguities are prepositional phrase (PP) attachment ambiguities and pronoun coreference ambiguities. In this paper, we discuss PatchComm, which uses commonsense knowledge to help resolve both kinds of ambiguities. To the best of our knowledge, we are the first to propose the general-purpose approach of using external commonsense knowledge bases to guide syntactic parsers. We evaluated PatchComm against the state-of-the-art (SOTA) spaCy parser on a PP attachment task and against the SOTA NeuralCoref module on a coreference task. Results show that PatchComm is successful at detecting syntactic ambiguities and using commonsense knowledge to help resolve them.},
author = {Xin, Yida and Lieberman, Henry and Chin, Peter},
booktitle = {Proceedings of the 18th International Conference on Principles of Knowledge Representation and Reasoning},
doi = {10.24963/kr.2021/75},
month = {November},
pages = {712–716},
title = {PATCHCOMM: Using Commonsense Knowledge to Guide Syntactic Parsers},
url = {https://doi.org/10.24963/kr.2021/75},
year = {2021}
}

Non-Volatile Memory Accelerated Geometric Multi-Scale Resolution Analysis

Cite

@inproceedings{wood_non-volatile_2021,
abstract = {Dimensionality reduction algorithms are standard tools in a researcher’s toolbox. Dimensionality reduction algorithms are frequently used to augment downstream tasks such as machine learning, data science, and also are exploratory methods for understanding complex phenomena. For instance, dimensionality reduction is commonly used in Biology as well as Neuroscience to understand data collected from biological subjects. However, dimensionality reduction techniques are limited by the von-Neumann architectures that they execute on. Specifically, data intensive algorithms such as dimensionality reduction techniques often require fast, high capacity, persistent memory which historically hardware has been unable to provide at the same time. In this paper, we present a re-implementation of an existing dimensionality reduction technique called Geometric Multi-Scale Resolution Analysis (GMRA) which has been accelerated via novel persistent memory technology called Memory Centric Active Storage (MCAS). Our implementation uses a specialized version of MCAS called PyMM that provides native support for Python datatypes including NumPy arrays and PyTorch tensors. We compare our PyMM implementation against a DRAM implementation, and show that when data fits in DRAM, PyMM offers competitive runtimes. When data does not fit in DRAM, our PyMM implementation is still able to process the data.},
author = {Wood, Andrew and Hershcovitch, Moshik and Waddington, Daniel and Cohen, Sarel and Wolf, Meredith and Suh, Hongjun and Zong, Weiyu and Chin, Peter},
booktitle = {2021 IEEE High Performance Extreme Computing Conference (HPEC)},
doi = {10.1109/HPEC49654.2021.9622854},
month = {September},
note = {ISSN: 2643-1971},
pages = {1–7},
title = {Non-Volatile Memory Accelerated Geometric Multi-Scale Resolution Analysis},
year = {2021}
}

Training Many-to-Many Recurrent Neural Networks with Target Propagation

Cite

@inproceedings{dai_training_2021,
abstract = {Deep neural networks trained with back-propagation have been the driving force for the progress in fields such as computer vision, natural language processing. However, back-propagation has often been criticized for its biological implausibility. More biologically plausible alternatives to backpropagation such as target propagation and feedback alignment have been proposed. But most of these learning algorithms are originally designed and tested for feedforward networks, and their ability for training recurrent networks and arbitrary computation graphs is not fully studied nor understood. In this paper, we propose a learning procedure based on target propagation for training multi-output recurrent networks. It opens doors to extending such biologically plausible models as general learning algorithms for arbitrary graphs.},
address = {Cham},
author = {Dai, Peilun and Chin, Sang},
booktitle = {Artificial Neural Networks and Machine Learning – ICANN 2021},
editor = {Farkaš, Igor and Masulli, Paolo and Otte, Sebastian and Wermter, Stefan},
isbn = {978-3-030-86380-7},
pages = {433–443},
publisher = {Springer International Publishing},
title = {Training Many-to-Many Recurrent Neural Networks with Target Propagation},
year = {2021}
}

RetroGAN: A Cyclic Post-Specialization System for Improving Out-of-Knowledge and Rare Word Representations

Cite

@inproceedings{colon-hernandez_retrogan_2021,
abstract = {Retrofitting is a technique used to move word vectors closer together or further apart in their space to reflect their relationships in a Knowledge Base (KB). However, retrofitting only works on concepts that are present in that KB. RetroGAN uses a pair of Generative Adversarial Networks (GANs) to learn a one-to-one mapping between concepts and their retrofitted counterparts. It applies that mapping (post-specializes) to handle concepts that do not appear in the original KB in a manner similar to how some natural language systems handle out-of-vocabulary entries. We test our system on three word-similarity benchmarks and a downstream sentence simplification task and achieve the state of the art (CARD-660). Altogether, our results demonstrate our system’s effectiveness for out-of-knowledge and rare word generalization.},
author = {Colon-Hernandez, Pedro and Xin, Yida and Lieberman, Henry and Havasi, Catherine and Breazeal, Cynthia and Chin, Peter},
booktitle = {Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021},
doi = {10.18653/v1/2021.findings-acl.183},
keywords = {Computer Science – Artificial Intelligence, Computer Science – Computation and Language},
note = {arXiv:2108.12941 [cs]},
pages = {2086–2095},
shorttitle = {RetroGAN},
title = {RetroGAN: A Cyclic Post-Specialization System for Improving Out-of-Knowledge and Rare Word Representations},
url = {http://arxiv.org/abs/2108.12941},
urldate = {2022-08-06},
year = {2021}
}

A Scale Invariant Measure of Flatness for Deep Network Minima

Cite

@inproceedings{rangamani_scale_2021,
abstract = {It has been empirically observed that the flatness of minima obtained from training deep networks seems to correlate with better generalization. However, for deep networks with positively homogeneous activations, most measures of flatness are not invariant to rescaling of the network parameters. This means that the measure of flatness can be made as small or as large as possible through rescaling, rendering the quantitative measures meaningless. In this paper we show that for deep networks with positively homogenous activations, these rescalings constitute equivalence relations, and that these equivalence relations induce a quotient manifold structure in the parameter space. Using an appropriate Riemannian metric, we propose a Hessian-based measure for flatness that is invariant to rescaling and perform simulations to empirically verify our claim. Finally we perform experiments to verify that our flatness measure correlates with generalization by using minibatch stochastic gradient descent with different batch sizes to find deep network minima with different generalization properties.},
author = {Rangamani, Akshay and Nguyen, Nam H. and Kumar, Abhishek and Phan, Dzung and Chin, Sang Peter and Tran, Trac D.},
booktitle = {ICASSP 2021 – 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
doi = {10.1109/ICASSP39728.2021.9413771},
month = {June},
note = {ISSN: 2379-190X},
pages = {1680–1684},
title = {A Scale Invariant Measure of Flatness for Deep Network Minima},
year = {2021}
}

Application of Seq2Seq Models on Code Correction

Cite

@article{huang_application_2021,
abstract = {We apply various seq2seq models on programming language correction tasks on Juliet Test Suite for C/C++ and Java of Software Assurance Reference Datasets and achieve 75% (for C/C++) and 56% (for Java) repair rates on these tasks. We introduce pyramid encoder in these seq2seq models, which significantly increases the computational efficiency and memory efficiency, while achieving similar repair rate to their nonpyramid counterparts. We successfully carry out error type classification task on ITC benchmark examples (with only 685 code instances) using transfer learning with models pretrained on Juliet Test Suite, pointing out a novel way of processing small programming language datasets.},
author = {Huang, Shan and Zhou, Xiao and Chin, Sang},
doi = {10.3389/frai.2021.590215},
issn = {2624-8212},
journal = {Frontiers in Artificial Intelligence},
month = {March},
pages = {590215},
title = {Application of Seq2Seq Models on Code Correction},
url = {https://www.frontiersin.org/articles/10.3389/frai.2021.590215/full},
urldate = {2022-08-06},
volume = {4},
year = {2021}
}

Revisiting the Prepositional-Phrase Attachment Problem Using Explicit Commonsense Knowledge

Publication:

Cite

@misc{xin_revisiting_2021,
abstract = {We revisit the challenging problem of resolving prepositional-phrase (PP) attachment ambiguity. To date, proposed solutions are either rule-based, where explicit grammar rules direct how to resolve ambiguities; or statistical, where the decision is learned from a corpus of labeled examples. We argue that explicit commonsense knowledge bases can provide an essential ingredient for making good attachment decisions. We implemented a module, named Patch-Comm, that can be used by a variety of conventional parsers, to make attachment decisions. Where the commonsense KB does not provide direct answers, we fall back on a more general system that infers “out-of-knowledge-base” assertions in a manner similar to the way some NLP systems handle out-of-vocabulary words. Our results suggest that the commonsense knowledge-based approach can provide the best of both worlds, integrating rule-based and statistical techniques. As the field is increasingly coming to recognize the importance of explainability in AI, a commonsense approach can enable NLP developers to better understand the behavior of systems, and facilitate natural dialogues with end users.},
author = {Xin, Yida and Lieberman, Henry and Chin, Peter},
keywords = {Computer Science – Artificial Intelligence, Computer Science – Computation and Language},
month = {February},
note = {arXiv:2102.00924 [cs]},
publisher = {arXiv},
title = {Revisiting the Prepositional-Phrase Attachment Problem Using Explicit Commonsense Knowledge},
url = {http://arxiv.org/abs/2102.00924},
urldate = {2022-08-06},
year = {2021}
}

Dreaming with ARC

Cite

@inproceedings{banburski_dreaming_2020,
abstract = {Current machine learning algorithms are highly specialized to whatever it is they are meant to do –– e.g. playing chess, picking up objects, or object recognition. How can we extend this to a system that could solve a wide range of problems? We argue that this can be achieved by a modular system –– one that can adapt to solving different problems by changing only the modules chosen and the order in which those modules are applied to the problem. The recently introduced ARC (Abstraction and Reasoning Corpus) dataset serves as an excellent test of abstract reasoning. Suited to the modular approach, the tasks depend on a set of human Core Knowledge inbuilt priors. In this paper we implement these priors as the modules of our system. We combine these modules using a neural-guided program synthesis.},
author = {Banburski, Andrzej and Gandhi, Anshula and Alford, Simon and Dandekar, Sylee and Chin, Sang and poggio, tomaso a},
booktitle = {Learning Meets Combinatorial Algorithms at NeurIPS2020},
title = {Dreaming with ARC},
url = {https://openreview.net/forum?id=-gjy2V1ko6t},
year = {2020}
}

NodeDrop: A Method for Finding Sufficient Network Architecture Size

Cite

@inproceedings{jensen_nodedrop_2020,
abstract = {Determining an appropriate number of features for each layer in a neural network is an important and difficult task. This task is especially important in applications on systems with limited memory or processing power. Many current approaches to reduce network size either utilize iterative procedures, which can extend training time significantly, or require very careful tuning of algorithm parameters to achieve reasonable results. In this paper we propose NodeDrop, a new method for eliminating features in a network. With NodeDrop, we define a condition to identify and guarantee which nodes carry no information, and then use regularization to encourage nodes to meet this condition. We find that NodeDrop drastically reduces the number of features in a network while maintaining high performance. NodeDrop reduces the number of parameters by a factor of 114x for a VGG like network on CIFAR10 without a drop in accuracy.},
author = {Jensen, Louis and Harer, Jacob and Chin, Sang},
booktitle = {2020 International Joint Conference on Neural Networks (IJCNN)},
doi = {10.1109/IJCNN48605.2020.9206880},
month = {July},
note = {ISSN: 2161-4407},
pages = {1–9},
title = {NodeDrop: A Method for Finding Sufficient Network Architecture Size},
year = {2020}
}

AdvMS: A Multi-Source Multi-Cost Defense Against Adversarial Attacks

Cite

@inproceedings{wang_advms_2020,
abstract = {Designing effective defense against adversarial attacks is a crucial topic as deep neural networks have been proliferated rapidly in many security-critical domains such as malware detection and self-driving cars. Conventional defense methods, although shown to be promising, are largely limited by their single-source single-cost nature: The robustness promotion tends to plateau when the defenses are made increasingly stronger while the cost tends to amplify. In this paper, we study principles of designing multi-source and multi-cost schemes where defense performance is boosted from multiple defending components. Based on this motivation, we propose a multi-source and multi-cost defense scheme, Adversarially Trained Model Switching (AdvMS), that inherits advantages from two leading schemes: adversarial training and random model switching. We show that the multi-source nature of AdvMS mitigates the performance plateauing issue and the multi-cost nature enables improving robustness at a flexible and adjustable combination of costs over different factors which can better suit specific restrictions and needs in practice.},
author = {Wang, Xiao and Wang, Siyue and Chen, Pin-Yu and Lin, Xue and Chin, Peter},
booktitle = {ICASSP 2020 – 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
doi = {10.1109/ICASSP40776.2020.9052967},
month = {May},
note = {ISSN: 2379-190X},
pages = {2902–2906},
title = {AdvMS: A Multi-Source Multi-Cost Defense Against Adversarial Attacks},
year = {2020}
}

GymFG: A Framework with a Gym Interface for FlightGear

Publication:

Cite

@misc{wood_gymfg_2020,
abstract = {Over the past decades, progress in deployable autonomous flight systems has slowly stagnated. This is reflected in today’s production air-crafts, where pilots only enable simple physics-based systems such as autopilot for takeoff, landing, navigation, and terrain/traffic avoidance. Evidently, autonomy has not gained the trust of the community where higher problem complexity and cognitive workload are required. To address trust, we must revisit the process for developing autonomous capabilities: modeling and simulation. Given the prohibitive costs for live tests, we need to prototype and evaluate autonomous aerial agents in a high fidelity flight simulator with autonomous learning capabilities applicable to flight systems: such a open-source development platform is not available. As a result, we have developed GymFG: GymFG couples and extends a high fidelity, open-source flight simulator and a robust agent learning framework to facilitate learning of more complex tasks. Furthermore, we have demonstrated the use of GymFG to train an autonomous aerial agent using Imitation Learning. With GymFG, we can now deploy innovative ideas to address complex problems and build the trust necessary to move prototypes to the real-world.},
author = {Wood, Andrew and Sydney, Ali and Chin, Peter and Thapa, Bishal and Ross, Ryan},
keywords = {Computer Science – Artificial Intelligence, Computer Science – Multiagent Systems, Computer Science – Robotics, I.2.1, I.6.5},
month = {April},
note = {arXiv:2004.12481 [cs]},
publisher = {arXiv},
shorttitle = {GymFG},
title = {GymFG: A Framework with a Gym Interface for FlightGear},
url = {http://arxiv.org/abs/2004.12481},
urldate = {2022-08-06},
year = {2020}
}

Block Switching: A Stochastic Approach for Deep Learning Security

Publication:

Cite

@misc{wang_block_2020,
abstract = {Recent study of adversarial attacks has revealed the vulnerability of modern deep learning models. That is, subtly crafted perturbations of the input can make a trained network with high accuracy produce arbitrary incorrect predictions, while maintain imperceptible to human vision system. In this paper, we introduce Block Switching (BS), a defense strategy against adversarial attacks based on stochasticity. BS replaces a block of model layers with multiple parallel channels, and the active channel is randomly assigned in the run time hence unpredictable to the adversary. We show empirically that BS leads to a more dispersed input gradient distribution and superior defense effectiveness compared with other stochastic defenses such as stochastic activation pruning (SAP). Compared to other defenses, BS is also characterized by the following features: (i) BS causes less test accuracy drop; (ii) BS is attack-independent and (iii) BS is compatible with other defenses and can be used jointly with others.},
annote = {Comment: Accepted by AdvML19: Workshop on Adversarial Learning Methods for Machine Learning and Data Mining at KDD, Anchorage, Alaska, USA, August 5th, 2019, 5 pages},
author = {Wang, Xiao and Wang, Siyue and Chen, Pin-Yu and Lin, Xue and Chin, Peter},
keywords = {Computer Science – Computer Vision and Pattern Recognition, Computer Science – Machine Learning},
month = {February},
note = {arXiv:2002.07920 [cs]},
publisher = {arXiv},
shorttitle = {Block Switching},
title = {Block Switching: A Stochastic Approach for Deep Learning Security},
url = {http://arxiv.org/abs/2002.07920},
urldate = {2022-08-06},
year = {2020}
}

A minimally invasive lens-free computational microendoscope

Cite

@article{shin_minimally_2019,
abstract = {A distal lensless microendoscope is developed to enable minimally invasive imaging with wide field of view and digital refocusing. Ultra-miniaturized microendoscopes are vital for numerous biomedical applications. Such minimally invasive imagers allow for navigation into hard-to-reach regions and observation of deep brain activity in freely moving animals. Conventional solutions use distal microlenses. However, as lenses become smaller and less invasive, they develop greater aberrations and restricted fields of view. In addition, most of the imagers capable of variable focusing require mechanical actuation of the lens, increasing the distal complexity and weight. Here, we demonstrate a distal lens-free approach to microendoscopy enabled by computational image recovery. Our approach is entirely actuation free and uses a single pseudorandom spatial mask at the distal end of a multicore fiber. Experimentally, this lensless approach increases the space-bandwidth product, i.e., field of view divided by resolution, by threefold over a best-case lens-based system. In addition, the microendoscope demonstrates color resolved imaging and refocusing to 11 distinct depth planes from a single camera frame without any actuated parts.},
author = {Shin, Jaewook and Tran, Dung N. and Stroud, Jasper R. and Chin, Sang and Tran, Trac D. and Foster, Mark A.},
doi = {10.1126/sciadv.aaw5595},
journal = {Science Advances},
note = {_eprint: https://www.science.org/doi/pdf/10.1126/sciadv.aaw5595},
number = {12},
pages = {eaaw5595},
title = {A minimally invasive lens-free computational microendoscope},
url = {https://www.science.org/doi/abs/10.1126/sciadv.aaw5595},
volume = {5},
year = {2019}
}

Optical coherence tomography using physical domain data compression to achieve MHz A-scan rates

Publication:

Cite

@article{stroud_optical_2019,
abstract = {The three-dimensional volumetric imaging capability of optical coherence tomography (OCT) leads to the generation of large amounts of data, which necessitates high speed acquisition followed by high dimensional image processing and visualization. This signal acquisition and processing pipeline demands high A-scan rates on the front end, which has driven researchers to push A-scan acquisition rates into the MHz regime. To this end, the optical time-stretch approach uses a mode locked laser (MLL) source, dispersion in optical fiber, and a single analog-to-digital converter (ADC) to achieve multi-MHz A-scan rates. While enabling impressive performance this Nyquist sampling approach is ultimately constrained by the sampling rate and bandwidth of the ADC. Additionally such an approach generates massive amounts of data. Here we present a compressed sensing (CS) OCT system that uses a MLL, electro-optic modulation, and optical dispersion to implement data compression in the physical domain and rapidly acquire real-time compressed measurements of the OCT signals. Compression in the analog domain prior to digitization allows for the use of lower bandwidth ADCs, which reduces cost and decreases the required data capacity of the sampling interface. By leveraging a compressive A-scan optical sampling approach and the joint sparsity of C-scan data we demonstrate 14.4-MHz to 144-MHz A-scan acquisition speeds using a sub-Nyquist 1.44 Gsample/sec ADC sampling rate. Furthermore we evaluate the impact of data compression and resulting imaging speed on image quality.},
author = {Stroud, Jasper R. and Liu, Luoluo and Chin, Sang and Tran, Trac D. and Foster, Mark A.},
doi = {10.1364/OE.27.036329},
journal = {Opt. Express},
keywords = {Image processing, Image quality, Optical coherence tomography, Three dimensional imaging, Mode locking, Fourier domain mode locking},
month = {December},
note = {Publisher: Optica Publishing Group},
number = {25},
pages = {36329–36339},
title = {Optical coherence tomography using physical domain data compression to achieve MHz A-scan rates},
url = {http://opg.optica.org/oe/abstract.cfm?URI=oe-27-25-36329},
volume = {27},
year = {2019}
}

Protecting Neural Networks with Hierarchical Random Switching: Towards Better Robustness-Accuracy Trade-off for Stochastic Defenses

Cite

@inproceedings{wang_protecting_2019,
abstract = {Despite achieving remarkable success in various domains, recent studies have uncovered the vulnerability of deep neural networks to adversarial perturbations, creating concerns on model generalizability and new threats such as prediction-evasive misclassification or stealthy reprogramming. Among different defense proposals, stochastic network defenses such as random neuron activation pruning or random perturbation to layer inputs are shown to be promising for attack mitigation. However, one critical drawback of current defenses is that the robustness enhancement is at the cost of noticeable performance degradation on legitimate data, e.g., large drop in test accuracy.This paper is motivated by pursuing for a better trade-off between adversarial robustness and test accuracy for stochastic network defenses. We propose Defense Efficiency Score (DES), a comprehensive metric that measures the gain in unsuccessful attack attempts at the cost of drop in test accuracy of any defense. To achieve a better DES, we propose hierarchical random switching (HRS), which protects neural networks through a novel randomization scheme. A HRS-protected model contains several blocks of randomly switching channels to prevent adversaries from exploiting fixed model structures and parameters for their malicious purposes. Extensive experiments show that HRS is superior in defending against state-of-the-art white-box and adaptive adversarial misclassification attacks. We also demonstrate the effectiveness of HRS in defending adversarial reprogramming, which is the first defense against adversarial programs. Moreover, in most settings the average DES of HRS is at least 5X higher than current stochastic network defenses, validating its significantly improved robustness-accuracy trade-off.},
author = {Wang, Xiao and Wang, Siyue and Chen, Pin-Yu and Wang, Yanzhi and Kulis, Brian and Lin, Xue and Chin, Sang},
booktitle = {Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence, IJCAI-19},
doi = {10.24963/ijcai.2019/833},
month = {July},
pages = {6013–6019},
publisher = {International Joint Conferences on Artificial Intelligence Organization},
title = {Protecting Neural Networks with Hierarchical Random Switching: Towards Better Robustness-Accuracy Trade-off for Stochastic Defenses},
url = {https://doi.org/10.24963/ijcai.2019/833},
year = {2019}
}

Tree-Transformer: A Transformer-Based Method for Correction of Tree-Structured Data

Publication:

Cite

@misc{harer_tree-transformer_2019,
abstract = {Many common sequential data sources, such as source code and natural language, have a natural tree-structured representation. These trees can be generated by fitting a sequence to a grammar, yielding a hierarchical ordering of the tokens in the sequence. This structure encodes a high degree of syntactic information, making it ideal for problems such as grammar correction. However, little work has been done to develop neural networks that can operate on and exploit tree-structured data. In this paper we present the Tree-Transformer \textemdash\ a novel neural network architecture designed to translate between arbitrary input and output trees. We applied this architecture to correction tasks in both the source code and natural language domains. On source code, our model achieved an improvement of \$25\%\$ \$\text\F\0.5\$ over the best sequential method. On natural language, we achieved comparable results to the most complex state of the art systems, obtaining a \$10\%\$ improvement in recall on the CoNLL 2014 benchmark and the highest to date \$\text\F\0.5\$ score on the AESW benchmark of \$50.43\$.},
author = {Harer, Jacob and Reale, Chris and Chin, Peter},
keywords = {Computer Science – Machine Learning, Statistics – Machine Learning, Computer Science – Computation and Language},
month = {August},
note = {arXiv:1908.00449 [cs, stat]},
publisher = {arXiv},
shorttitle = {Tree-Transformer},
title = {Tree-Transformer: A Transformer-Based Method for Correction of Tree-Structured Data},
url = {http://arxiv.org/abs/1908.00449},
urldate = {2022-08-06},
year = {2019}
}

JOBS: Joint-Sparse Optimization from Bootstrap Samples

Cite

@inproceedings{liu_jobs_2019,
abstract = {Classical sparse regression based on ℓ1 minimization solves the least squares problem with all available measurements via sparsity-promoting regularization. In challenging practical applications with high levels of noise and missing or adversarial samples, solving the problem using all measurements simultaneously may fail. In this paper, we propose a robust global sparse recovery strategy, named JOBS, which uses bootstrap samples of measurements to improve sparse regression in difficult cases. K measurement vectors are generated from the original pool of m measurements via bootstrapping, with each bootstrap sample containing L elements, and then a joint-sparse constraint is enforced to ensure support consistency among multiple predictors. The final estimate is obtained by averaging over K estimators. The performance limits associated with finite bootstrap sampling ratio L/m and number of estimates K is analyzed theoretically. Simulation results validate the theoretical analysis of proper choice of (L,K) and show that the proposed method yields state-of-the-art recovery performance, outperforming ℓ1 minimization and other existing bootstrap-based techniques, especially when the number of measurements are limited. With a proper choice of bootstrap sampling ratio (0.3-0.5) and a reasonably large number of estimates K (≥ 30), the SNR improvement over the baseline ℓ1-minimization algorithm can reach up to 336%.},
author = {Liu, Luoluo and Chin, Sang Peter and Tran, Trac D.},
booktitle = {2019 IEEE International Symposium on Information Theory (ISIT)},
doi = {10.1109/ISIT.2019.8849654},
month = {July},
note = {ISSN: 2157-8117},
pages = {2689–2693},
title = {JOBS: Joint-Sparse Optimization from Bootstrap Samples},
year = {2019}
}

Deep Learning for Minimal-context Block Tracking through Side-channel Analysis

Cite

@inproceedings{jensen_deep_2019,
abstract = {It is well known that electromagnetic and power side-channel attacks allow extraction of unintended information from a computer processor. However, little work has been done to quantify how small a sample is needed in order to glean meaningful information about a program’s execution. This paper quantifies this minimum context by training a deep-learning model to track and classify program block types given small windows of side-channel data. We show that a window containing approximately four clock cycles suffices to predict block type with our experimental setup. This implies a high degree of information leakage through side channels, allowing for the external monitoring of embedded systems and Internet of Things devices.},
author = {Jensen, L. and Brown, G. and Wang, X. and Harer, J. and Chin, S.},
booktitle = {ICASSP 2019 – 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
doi = {10.1109/ICASSP.2019.8683845},
month = {May},
note = {ISSN: 2379-190X},
pages = {3207–3211},
title = {Deep Learning for Minimal-context Block Tracking through Side-channel Analysis},
year = {2019}
}

Reducing Sampling Ratios and Increasing Number of Estimates Improve Bagging in Sparse Regression

Cite

@inproceedings{liu_reducing_2019,
abstract = {Bagging, a powerful ensemble method from machine learning, has shown the ability to improve the performance of unstable predictors in difficult practical settings. Although Bagging is most well-known for its application in classification problems, here we demonstrate that employing Bagging in sparse regression improves performance compared to the baseline method (I1 minimization). Although the original Bagging method uses a bootstrap sampling ratio of 1, such that the sizes of the bootstrap samples L are the same as the total number of data points m, we generalize the bootstrap sampling ratio to explore the optimal sampling ratios for various cases. The performance limits associated with different choices of bootstrap sampling ratio L/m and number of estimates K are analyzed theoretically. Simulation results show that a lower L/m ratio (0.6 – 0.9) leads to better performance than the conventional choice (L/m = 1), especially in challenging cases with low levels of measurements. With the reduced sampling rate, SNR improves over the original Bagging method by up to 24% and over the base algorithm I1 minimization by up to 367%. With a properly chosen sampling ratio, a reasonably small number of estimates (K = 30) gives a satisfying result, although increasing K is discovered to always improve or at least maintain performance.},
author = {Liu, Luoluo and Chin, Sang Peter and Tran, Trac D.},
booktitle = {2019 53rd Annual Conference on Information Sciences and Systems (CISS)},
doi = {10.1109/CISS.2019.8692865},
month = {March},
pages = {1–5},
title = {Reducing Sampling Ratios and Increasing Number of Estimates Improve Bagging in Sparse Regression},
year = {2019}
}

Learning to Repair Software Vulnerabilities with Generative Adversarial Networks

Cite

@inproceedings{harer_learning_2018,
abstract = {Motivated by the problem of automated repair of software vulnerabilities, we propose an adversarial learning approach that maps from one discrete source domain to another target domain without requiring paired labeled examples or source and target domains to be bijections. We demonstrate that the proposed adversarial learning approach is an effective technique for repairing software vulnerabilities, performing close to seq2seq approaches that require labeled pairs. The proposed Generative Adversarial Network approach is application-agnostic in that it can be applied to other problems similar to code repair, such as grammar correction or sentiment translation.},
address = {Red Hook, NY, USA},
author = {Harer, Jacob A. and Ozdemir, Onur and Lazovich, Tomo and Reale, Christopher P. and Russell, Rebecca L. and Kim, Louis Y. and Chin, Peter},
booktitle = {Proceedings of the 32nd International Conference on Neural Information Processing Systems},
note = {event-place: Montréal, Canada},
pages = {7944–7954},
publisher = {Curran Associates Inc.},
series = {NIPS’18},
title = {Learning to Repair Software Vulnerabilities with Generative Adversarial Networks},
year = {2018}
}

RECONSTRUCTION-FREE DEEP CONVOLUTIONAL NEURAL NETWORKS FOR PARTIALLY OBSERVED IMAGES

Cite

@inproceedings{nair_reconstruction-free_2018,
abstract = {Conventional image discrimination tasks are performed on fully observed images. In challenging real imaging scenarios, where sensing systems are energy demanding or need to operate with limited bandwidth and exposure-time budgets, or defective pixels, where the data collected often suffers from missing information, and this makes the task extremely hard. In this paper, we leverage Convolutional Neural Networks (CNNs) to extract information from partially observed images. While pre-trained CNNs fail significantly even with such a small percentage of the input missing, our proposed framework demonstrates the ability to overcome it after training on fully-observed and partially-observed images at a few observation ratios. We demonstrate that our method is indeed reconstruction-free, retraining-free and generalizable to previously untrained-on observation ratios and it remains effective in two different visual tasks – image classification and object detection. Our framework performs well even for test images with only 10% of pixels available and outperforms the reconstruct-then-classify pipeline in these challenging scenarios for small observation fractions.},
author = {Nair, Arun and Liu, Luoluo and Rangamani, Akshay and Chin, Peter and Lediju Bell, Muyinatu A. and Tran, Trac D.},
booktitle = {2018 IEEE Global Conference on Signal and Information Processing (GlobalSIP)},
doi = {10.1109/GlobalSIP.2018.8646498},
month = {November},
pages = {400–404},
title = {RECONSTRUCTION-FREE DEEP CONVOLUTIONAL NEURAL NETWORKS FOR PARTIALLY OBSERVED IMAGES},
year = {2018}
}

Defensive Dropout for Hardening Deep Neural Networks under Adversarial Attacks

Cite

@inproceedings{wang_defensive_2018,
abstract = {Deep neural networks (DNNs) are known vulnerable to adversarial attacks. That is, adversarial examples, obtained by adding delicately crafted distortions onto original legal inputs, can mislead a DNN to classify them as any target labels. This work provides a solution to hardening DNNs under adversarial attacks through defensive dropout. Besides using dropout during training for the best test accuracy, we propose to use dropout also at test time to achieve strong defense effects. We consider the problem of building robust DNNs as an attacker-defender two-player game, where the attacker and the defender know each others’ strategies and try to optimize their own strategies towards an equilibrium. Based on the observations of the effect of test dropout rate on test accuracy and attack success rate, we propose a defensive dropout algorithm to determine an optimal test dropout rate given the neural network model and the attacker’s strategy for generating adversarial examples. We also investigate the mechanism behind the outstanding defense effects achieved by the proposed defensive dropout. Comparing with stochastic activation pruning (SAP), another defense method through introducing randomness into the DNN model, we find that our defensive dropout achieves much larger variances of the gradients, which is the key for the improved defense effects (much lower attack success rate). For example, our defensive dropout can reduce the attack success rate from 100% to 13.89% under the currently strongest attack i.e., C&W attack on MNIST dataset.},
author = {Wang, Siyue and Wang, Xiao and Zhao, Pu and Wen, Wujie and Kaeli, David and Chin, Peter and Lin, Xue},
booktitle = {2018 IEEE/ACM International Conference on Computer-Aided Design (ICCAD)},
doi = {10.1145/3240765.3264699},
month = {November},
note = {ISSN: 1558-2434},
pages = {1–8},
title = {Defensive Dropout for Hardening Deep Neural Networks under Adversarial Attacks},
year = {2018}
}

Automated software vulnerability detection with machine learning

Publication:

Cite

@misc{harer_automated_2018,
abstract = {Thousands of security vulnerabilities are discovered in production software each year, either reported publicly to the Common Vulnerabilities and Exposures database or discovered internally in proprietary code. Vulnerabilities often manifest themselves in subtle ways that are not obvious to code reviewers or the developers themselves. With the wealth of open source code available for analysis, there is an opportunity to learn the patterns of bugs that can lead to security vulnerabilities directly from data. In this paper, we present a data-driven approach to vulnerability detection using machine learning, specifically applied to C and C++ programs. We first compile a large dataset of hundreds of thousands of open-source functions labeled with the outputs of a static analyzer. We then compare methods applied directly to source code with methods applied to artifacts extracted from the build process, finding that source-based models perform better. We also compare the application of deep neural network models with more traditional models such as random forests and find the best performance comes from combining features learned by deep models with tree-based models. Ultimately, our highest performing model achieves an area under the precision-recall curve of 0.49 and an area under the ROC curve of 0.87.},
author = {Harer, Jacob A. and Kim, Louis Y. and Russell, Rebecca L. and Ozdemir, Onur and Kosta, Leonard R. and Rangamani, Akshay and Hamilton, Lei H. and Centeno, Gabriel I. and Key, Jonathan R. and Ellingwood, Paul M. and Antelman, Erik and Mackay, Alan and McConley, Marc W. and Opper, Jeffrey M. and Chin, Peter and Lazovich, Tomo},
keywords = {Computer Science – Machine Learning, Statistics – Machine Learning, Computer Science – Software Engineering},
month = {August},
note = {arXiv:1803.04497 [cs, stat]},
publisher = {arXiv},
title = {Automated software vulnerability detection with machine learning},
url = {http://arxiv.org/abs/1803.04497},
urldate = {2022-08-06},
year = {2018}
}

Sparse Coding and Autoencoders

Cite

@inproceedings{rangamani_sparse_2018,
abstract = {In this work we study the landscape of squared loss of an Autoencoder when the data generative model is that of “Sparse Coding”/“Dictionary Learning”. The neural net considered is an \$\mathbbR\textasciicircumn\rightarrow \mathbbR\textasciicircumn\$ mapping and has a single ReLU activation layer of size \$h \textgreater n\$. The net has access to vectors \$yın \mathbbR\textasciicircumn\$ obtained as \$y=A\textasciicircum\astx\textasciicircum\ast\$ where \$x\textasciicircum\astın \mathbbR\textasciicircumh\$ are sparse high dimensional vectors and \$A\textasciicircum\astın \mathbbR\textasciicircumn\times h$^\textrm\ast\$$, we prove that the norm of the expected gradient of the squared loss function is asymptotically (in sparse code dimension) negligible for all points in a small neighborhood of $^\textrm\ast\$$. This is supported with experimental evidence using synthetic data. We conduct experiments to suggest that $^\textrm\ast\$$ sits at the bottom of a well in the landscape and we also give experiments showing that gradient descent on this loss function gets columnwise very close to the original dictionary even with far enough initialization. Along the way we prove that a layer of ReLU gates can be set up to automatically recover the support of the sparse codes. Since this property holds independent of the loss function we believe that it could be of independent interest. A full version of this paper is accessible at: https://arxiv.org/abs/1708.03735},
author = {Rangamani, Akshay and Mukherjee, Anirbit and Basu, Amitabh and Arora, Ashish and Ganapathi, Tejaswini and Chin, Sang and Tran, Trac D.},
booktitle = {2018 IEEE International Symposium on Information Theory (ISIT)},
doi = {10.1109/ISIT.2018.8437533},
month = {June},
note = {ISSN: 2157-8117},
pages = {36–40},
title = {Sparse Coding and Autoencoders},
year = {2018}
}

Widefield compressive multiphoton microscopy

Publication:

Cite

@article{alemohammad_widefield_2018,
abstract = {A single-pixel compressively sensed architecture is exploited to simultaneously achieve a 10× reduction in acquired data compared with the Nyquist rate, while alleviating limitations faced by conventional widefield temporal focusing microscopes due to scattering of the fluorescence signal. Additionally, we demonstrate an adaptive sampling scheme that further improves the compression and speed of our approach.},
author = {Alemohammad, Milad and Shin, Jaewook and Tran, Dung N. and Stroud, Jasper R. and Chin, Sang Peter and Tran, Trac D. and Foster, Mark A.},
doi = {10.1364/OL.43.002989},
journal = {Opt. Lett.},
keywords = {Digital micromirror devices, Image processing, Image quality, Computational imaging, Adaptive imaging, Imaging through turbid media, Multiphoton microscopy, Nonlinear microscopy, Three dimensional imaging, Two photon imaging},
month = {June},
note = {Publisher: Optica Publishing Group},
number = {12},
pages = {2989–2992},
title = {Widefield compressive multiphoton microscopy},
url = {http://opg.optica.org/ol/abstract.cfm?URI=ol-43-12-2989},
volume = {43},
year = {2018}
}

Using Deep Learning to Extract Scenery Information in Real Time Spatiotemporal Compressed Sensing

Cite

@inproceedings{wang_using_2018,
abstract = {One of the problems of real time compressed sensing system is the computational cost of the reconstruction algorithms. It is especially problematic for close loop sensory applications where the sensory parameters needs to be constantly adjust to adapt to a dynamic scene. Through a preliminary experiment with MNIST dataset, we showed that we can extract some scene information (object recognition, scene movement direction and speed) based on the compressed samples using a deep convolutional neural network. It achieves 100% percent accuracy in distinguishing moving velocity, 96.22% in recognizing the digit and 90.04% in detecting moving direction after the code images are re-centered. Even though the classification accuracy drops slightly compared to using original videos, the computational speed is two time faster than classification on videos directly. This method also eliminates the need for sparse reconstruction prior to classification.},
author = {Wang, Xiao and Zhang, Jie and Xiong, Tao and Tran, Trac Duy and Chin, Sang Peter and Etienne-Cummings, Ralph},
booktitle = {2018 IEEE International Symposium on Circuits and Systems (ISCAS)},
doi = {10.1109/ISCAS.2018.8351736},
month = {May},
note = {ISSN: 2379-447X},
pages = {1–4},
title = {Using Deep Learning to Extract Scenery Information in Real Time Spatiotemporal Compressed Sensing},
year = {2018}
}

Tree Structured Multimedia Signal Modeling

Cite

@inproceedings{ma_tree_2018,
abstract = {Current solutions to multimedia modeling tasks feature sequential models and static tree-structured models. Sequential models, especially models based on Bidirectional LSTM (BLSTM) and Multilayer LSTM networks, have been widely applied on video, sound, music and text corpora. Despite their success in achieving state-of-the-art results on several multimedia processing tasks, sequential models always fail to emphasize short-term dependency relations, which are crucial in most sequential multimedia data. Tree-structured models are able to overcome this defect. The static tree-structured LSTM presented by Tai et al. (Tai, Socher, and Manning 2015) forcingly breaks down the dependencies between elements in each semantic group and those outside the group, while preserves chain-dependencies among semantic groups and among nodes in the same group. Though the tree-LSTM network is able to better represent the dependency structure of multimedia data, it requires the dependency relations of the input data to be known before it is fed into the network. This is hard to achieve since for most types of multimedia data there exists no parsers which can detect the dependency structure of every input sequence accurately enough. In order to preserve dependency information while eliminating the necessity of a perfect parser, in this paper we present a novel neural network architecture which 1) is self-expandable and 2) maintains the layered dependency structure of incoming multimedia data. We call our new neural network architecture Seq2Tree network. A Seq2Tree model is applicable on classification, prediction and generation tasks with task-specific adjustments of the model. We prove by experiments that our Seq2Tree model performs well in all the three types of tasks.},
author = {Ma, Weicheng and Cao, Kai and Li, Xiang and Chin, Peter},
booktitle = {Proceedings of the Thirty-First International Florida Artificial Intelligence Research Society Conference, FLAIRS 2018, Melbourne, Florida, USA. May 21-23 2018},
editor = {Brawner, Keith and Rus, Vasile},
pages = {184–188},
publisher = {AAAI Press},
title = {Tree Structured Multimedia Signal Modeling},
url = {https://aaai.org/ocs/index.php/FLAIRS/FLAIRS18/paper/view/17675},
year = {2018}
}

Deep learning-based classification and anomaly detection of side-channel signals

Cite

@inproceedings{wang_deep_2018,
abstract = {In computer systems, information leaks from the physical hardware through side-channel signals such as power draw. We can exploit these signals to infer the state of ongoing computational tasks without having direct access to the device. This paper investigates the application of recent deep learning techniques to side-channel analysis in both classification of machine state and anomaly detection. We use real data collected from three different devices: an Arduino, a Raspberry Pi, and a Siemens PLC. For classification we compare the performance of a Multi-Layer Perceptron and a Long Short-Term Memory classifiers. Both achieve near-perfect accuracy on binary classification and around 90% accuracy on a multi-class problem. For anomaly detection we explore an autoencoder based model. Our experiments show the potential of using these deep learning techniques in side-channel analysis and cyber-attack detection.},
author = {Wang, Xiao and Zhou, Quan and Harer, Jacob and Brown, Gavin and Qiu, Shangran and Dou, Zhi and Wang, John and Hinton, Alan and Gonzalez, Carlos Aguayo and Chin, Peter},
booktitle = {Cyber Sensing 2018},
doi = {10.1117/12.2311329},
editor = {Ternovskiy, Igor V. and Chin, Peter},
keywords = {Anomaly Detection, Classification, Deep Learning, Side-channel Analysis},
note = {Backup Publisher: International Society for Optics and Photonics},
pages = {37 — 44},
publisher = {SPIE},
title = {Deep learning-based classification and anomaly detection of side-channel signals},
url = {https://doi.org/10.1117/12.2311329},
volume = {10630},
year = {2018}
}

Sound Signal Processing with Seq2Tree Network

Cite

@inproceedings{ma_sound_2018,
abstract = {Long Short-Term Memory (LSTM) and its variants have been the standard solution to sequential data processing tasks because of their
ability to preserve previous information weighted on distance. This feature provides the LSTM family with additional information
in predictions, compared to regular Recurrent Neural Networks (RNNs) and Bag-of-Words (BOW) models. In other words, LSTM
networks assume the data to be chain-structured. The longer the distance between two data points, the less related the data points are.
However, this is usually not the case for real multimedia signals including text, sound and music. In real data, this chain-structured
dependency exists only across meaningful groups of data units but not over single units directly. For example, in a prediction task over
sound signals, a meaningful word could give a strong hint to its following word as a whole but not the first phoneme of that word.
This undermines the ability of LSTM networks in modeling multimedia data, which is pattern-rich. In this paper we take advantage of
Seq2Tree network, a dynamically extensible tree-structured neural network architecture which helps solve the problem LSTM networks
face in sound signal processing tasks—the unbalanced connections among data units inside and outside semantic groups. Experiments
show that Seq2Tree network outperforms the state-of-the-art Bidirectional LSTM (BLSTM) model on a signal and noise separation task
(CHiME Speech Separation and Recognition Challenge).},
author = {Ma, Weicheng and Cao, Kai and Ni, Zhaoheng and Chin, Peter and Li, Xiang},
booktitle = {Proceedings of the Eleventh International Conference on Language Resources and Evaluation, LREC 2018, Miyazaki, Japan, May 7-12, 2018},
editor = {Calzolari, Nicoletta and Choukri, Khalid and Cieri, Christopher and Declerck, Thierry and Goggi, Sara and Hasida, Kôiti and Isahara, Hitoshi and Maegaard, Bente and Mariani, Joseph and Mazo, Hélène and Moreno, Asunción and Odijk, Jan and Piperidis, Stelios and Tokunaga, Takenobu},
publisher = {European Language Resources Association (ELRA)},
title = {Sound Signal Processing with Seq2Tree Network},
url = {http://www.lrec-conf.org/proceedings/lrec2018/summaries/164.html},
year = {2018}
}

An Unsupervised Compressed Sensing Algorithm for Multi-Channel Neural Recording and Spike Sorting

Cite

@article{xiong_unsupervised_2018,
abstract = {We propose an unsupervised compressed sensing (CS)-based framework to compress, recover, and cluster neural action potentials. This framework can be easily integrated into high-density multi-electrode neural recording VLSI systems. Embedding spectral clustering and group structures in dictionary learning, we extend the proposed framework to unsupervised spike sorting without prior label information. Additionally, we incorporate group sparsity concepts in the dictionary learning to enable the framework for multi-channel neural recordings, as in tetrodes. To further improve spike sorting success rates in the CS framework, we embed template matching in sparse coding to jointly predict clusters of spikes. Our experimental results demonstrate that the proposed CS-based framework can achieve a high compression ratio (8:1 to 20:1), with a high quality reconstruction performance (\textgreater8 dB) and a high spike sorting accuracy (\textgreater90%).},
author = {Xiong, Tao and Zhang, Jie and Martinez-Rubio, Clarissa and Thakur, Chetan S. and Eskandar, Emad N. and Chin, Sang Peter and Etienne-Cummings, Ralph and Tran, Trac D.},
doi = {10.1109/TNSRE.2018.2830354},
issn = {1558-0210},
journal = {IEEE Transactions on Neural Systems and Rehabilitation Engineering},
month = {June},
number = {6},
pages = {1121–1130},
title = {An Unsupervised Compressed Sensing Algorithm for Multi-Channel Neural Recording and Spike Sorting},
volume = {26},
year = {2018}
}

A Greedy Pursuit Algorithm for Separating Signals from Nonlinear Compressive Observations

Cite

@inproceedings{chin_greedy_2018,
abstract = {In this paper we study the unmixing problem which aims to separate a set of structured signals from their superposition. In this paper, we consider the scenario in which the mixture is observed via nonlinear compressive measurements. We present a fast, robust, greedy algorithm called Unmixing Matching Pursuit (UnmixMP) to solve this problem. We prove rigorously that the algorithm can recover the constituents from their noisy nonlinear compressive measurements with arbitrarily small error. We compare our algorithm to the Demixing with Hard Thresholding (DHT) algorithm [1], in a number of experiments on synthetic and real data.},
author = {Chin, Sang and Tran, Trac D. and Tran, Dung and Rangamani, Akshay},
booktitle = {2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
doi = {10.1109/ICASSP.2018.8462387},
month = {April},
note = {ISSN: 2379-190X},
pages = {2171–2175},
title = {A Greedy Pursuit Algorithm for Separating Signals from Nonlinear Compressive Observations},
year = {2018}
}

Domain specific inpainting with concurrently pretrained generative adversarial networks

Cite

@inproceedings{zhou_domain_2017,
abstract = {This paper works on recovering noised or cropped images by training a neural network based on DCGAN’s and WGAN’s structures. Various approaches have been used including the generative model using deep neural network, modified inputs for the generator, application of pre-trained classification model, etc. In this paper, the advantages of different GANs are combined to tackle the image inpainting problem. Besides, this paper proposed a special data feeding approach which concurrently trains generator and discriminator of a GAN to further improve the performance of domain specific inpainting tasks. This architecture is evaluated and tested on LSUN[1] dataset with two different domains. The results reflect the feasibility of the approach, and comparing to the existing semantic inpainting methods, this architecture further improves both numerical loss and classification accuracy.},
author = {Zhou, Xiao and Wang, Chengchen and Xu, Yiteng and Wang, Xiao and Chin, Peter},
booktitle = {2017 IEEE Global Conference on Signal and Information Processing (GlobalSIP)},
doi = {10.1109/GlobalSIP.2017.8309148},
month = {November},
pages = {1185–1189},
title = {Domain specific inpainting with concurrently pretrained generative adversarial networks},
year = {2017}
}

Deep Image-to-Image Recurrent Network with Shape Basis Learning for Automatic Vertebra Labeling in Large-Scale 3D CT Volumes

Cite

@inproceedings{yang_deep_2017,
abstract = {Automatic vertebra localization and identification in 3D medical images plays an important role in many clinical tasks, including pathological diagnosis, surgical planning and postoperative assessment. In this paper, we propose an automatic and efficient algorithm to localize and label the vertebra centroids in 3D CT volumes. First, a deep image-to-image network (DI2IN) is deployed to initialize vertebra locations, employing the convolutional encoder-decoder architecture. Next, the centroid probability maps from DI2IN are modeled as a sequence according to the spatial relationship of vertebrae, and evolved with the convolutional long short-term memory (ConvLSTM) model. Finally, the landmark positions are further refined and regularized by another neural network with a learned shape basis. The whole pipeline can be conducted in the end-to-end manner. The proposed method outperforms other state-of-the-art methods on a public database of 302 spine CT volumes with various pathologies. To further boost the performance and validate that large labeled training data can benefit the deep learning algorithms, we leverage the knowledge of additional 1000 3D CT volumes from different patients. Our experimental results show that training with a large database improves the performance of proposed framework by a large margin and achieves an identification rate of 89%.},
address = {Cham},
author = {Yang, Dong and Xiong, Tao and Xu, Daguang and Zhou, S. Kevin and Xu, Zhoubing and Chen, Mingqing and Park, JinHyeong and Grbic, Sasa and Tran, Trac D. and Chin, Sang Peter and Metaxas, Dimitris and Comaniciu, Dorin},
booktitle = {Medical Image Computing and Computer Assisted Intervention − MICCAI 2017},
editor = {Descoteaux, Maxime and Maier-Hein, Lena and Franz, Alfred and Jannin, Pierre and Collins, D. Louis and Duchesne, Simon},
isbn = {978-3-319-66179-7},
pages = {498–506},
publisher = {Springer International Publishing},
title = {Deep Image-to-Image Recurrent Network with Shape Basis Learning for Automatic Vertebra Labeling in Large-Scale 3D CT Volumes},
year = {2017}
}

Spatiotemporal compressed sensing for video compression

Cite

@inproceedings{xiong_spatiotemporal_2017,
abstract = {We present a hardware-friendly spatiotemporal compressed sensing framework for video compression. The spatiotemporal compressed sensing incorporates random sampling in both spatial and temporal domain to encode the video scene into a single coded image. During decoding, the video is reconstructed using dictionary learning and sparse recovery. The evaluation results demonstrate the proposed approach can achieve high compression rate (10 : 1-30 : 1) and robustness reconstruction quality (\textgreater 20dB) on noisy database. Additionally, it also enables power efficient and real-time CMOS implementation (0.7 nJ/pixel).},
author = {Xiong, Tao and Rattray, John and Zhang, Jie and Thakur, Chetan Singh and Chin, Sang Peter and Tran, Trac D. and Etienne-Cummings, Ralph},
booktitle = {2017 IEEE 60th International Midwest Symposium on Circuits and Systems (MWSCAS)},
doi = {10.1109/MWSCAS.2017.8052917},
month = {August},
note = {ISSN: 1558-3899},
pages = {289–292},
title = {Spatiotemporal compressed sensing for video compression},
year = {2017}
}

Live demonstration: A compact all-CMOS spatiotemporal compressed sensing video camera

Cite

@inproceedings{xiong_live_2017,
abstract = {A compact all-CMOS spatiotemporal compressed sensing (CS) video camera is demonstrated. This CS-based framework [1], implemented on integrated circuits, is able to achieve 20-fold reduction in the readout speed and consumes only 14μW to provide 100 fps videos. Taking advantage of dictionary learning and sparse recovery, this prototype image sensor (127×90 pixels) can reconstruct 100 fps videos from the coded images sampled at 5 fps.},
author = {Xiong, Tao and Zhang, Jie and Thakur, Chetan Singh and Rattray, John and Chin, Sang Peter and Tran, Trac D. and Etienne-Cummings, Ralph},
booktitle = {2017 IEEE International Symposium on Circuits and Systems (ISCAS)},
doi = {10.1109/ISCAS.2017.8050405},
month = {May},
note = {ISSN: 2379-447X},
pages = {1–1},
title = {Live demonstration: A compact all-CMOS spatiotemporal compressed sensing video camera},
year = {2017}
}

Automatic Vertebra Labeling in Large-Scale 3D CT Using Deep Image-to-Image Network with Message Passing and Sparsity Regularization

Cite

@inproceedings{yang_automatic_2017,
abstract = {Automatic localization and labeling of vertebra in 3D medical images plays an important role in many clinical tasks, including pathological diagnosis, surgical planning and postoperative assessment. However, the unusual conditions of pathological cases, such as the abnormal spine curvature, bright visual imaging artifacts caused by metal implants, and the limited field of view, increase the difficulties of accurate localization. In this paper, we propose an automatic and fast algorithm to localize and label the vertebra centroids in 3D CT volumes. First, we deploy a deep image-to-image network (DI2IN) to initialize vertebra locations, employing the convolutional encoder-decoder architecture together with multi-level feature concatenation and deep supervision. Next, the centroid probability maps from DI2IN are iteratively evolved with the message passing schemes based on the mutual relation of vertebra centroids. Finally, the localization results are refined with sparsity regularization. The proposed method is evaluated on a public dataset of 302 spine CT volumes with various pathologies. Our method outperforms other state-of-the-art methods in terms of localization accuracy. The run time is around 3 seconds on average per case. To further boost the performance, we retrain the DI2IN on additional 1000+ 3D CT volumes from different patients. To the best of our knowledge, this is the first time more than 1000 3D CT volumes with expert annotation are adopted in experiments for the anatomic landmark detection tasks. Our experimental results show that training with such a large dataset significantly improves the performance and the overall identification rate, for the first time by our knowledge, reaches 90%.},
address = {Cham},
author = {Yang, Dong and Xiong, Tao and Xu, Daguang and Huang, Qiangui and Liu, David and Zhou, S. Kevin and Xu, Zhoubing and Park, JinHyeong and Chen, Mingqing and Tran, Trac D. and Chin, Sang Peter and Metaxas, Dimitris and Comaniciu, Dorin},
booktitle = {Information Processing in Medical Imaging},
editor = {Niethammer, Marc and Styner, Martin and Aylward, Stephen and Zhu, Hongtu and Oguz, Ipek and Yap, Pew-Thian and Shen, Dinggang},
isbn = {978-3-319-59050-9},
pages = {633–644},
publisher = {Springer International Publishing},
title = {Automatic Vertebra Labeling in Large-Scale 3D CT Using Deep Image-to-Image Network with Message Passing and Sparsity Regularization},
year = {2017}
}

Automatic similarity detection and clustering of data

Cite

@inproceedings{einstein_automatic_2017,
abstract = {An algorithm was created which identifies the number of unique clusters in a dataset and assigns the data to the clusters. A cluster is defined as a group of data which share similar characteristics. Similarity is measured using the dot product between two vectors where the data are input as vectors. Unlike other clustering algorithms such as K-means, no knowledge of the number of clusters is required. This allows for an unbiased analysis of the data. The automatic cluster detection algorithm (ACD), is executed in two phases: an averaging phase and a clustering phase. In the averaging phase, the number of unique clusters is detected. In the clustering phase, data are matched to the cluster to which they are most similar. The ACD algorithm takes a matrix of vectors as an input and outputs a 2D array of the clustered data. The indices of the output correspond to a cluster, and the elements in each cluster correspond to the position of the datum in the dataset. Clusters are vectors in N-dimensional space, where N is the length of the input vectors which make up the matrix. The algorithm is distributed, increasing computational efficiency},
author = {Einstein, Craig and Chin, Peter},
booktitle = {Cyber Sensing 2017},
doi = {10.1117/12.2267844},
editor = {Ternovskiy, Igor V. and Chin, Peter},
note = {Backup Publisher: International Society for Optics and Photonics},
pages = {144 — 150},
publisher = {SPIE},
title = {Automatic similarity detection and clustering of data},
url = {https://doi.org/10.1117/12.2267844},
volume = {10185},
year = {2017}
}

A provable nonconvex model for factoring nonnegative matrices

Cite

@inproceedings{tran_provable_2017,
abstract = {We study the Nonnegative Matrix Factorization problem which approximates a nonnegative matrix by a low-rank factorization. This problem is particularly important in Machine Learning, and finds itself in a large number of applications. Unfortunately, the original formulation is ill-posed and NP-hard. In this paper, we propose a row sparse model based on Row Entropy Minimization to solve the NMF problem under separable assumption which states that each data point is a convex combination of a few distinct data columns. We utilize the concentration of the entropy function and the ℓ∞ norm to concentrate the energy on the least number of latent variables. We prove that under the separability assumption, our proposed model robustly recovers data columns that generate the dataset, even when the data is corrupted by noise. We empirically justify the robustness of the proposed model and show that it is significantly more robust than the state-of-the-art separable NMF algorithms.},
author = {Tran, Dung N. and Chin, Sang P. and Tran, Trac D.},
booktitle = {2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
doi = {10.1109/ICASSP.2017.7952559},
month = {March},
note = {ISSN: 2379-190X},
pages = {2262–2266},
title = {A provable nonconvex model for factoring nonnegative matrices},
year = {2017}
}

Topological and statistical behavior classifiers for tracking applications

Cite

@article{bendich_topological_2016,
abstract = {This paper introduces a method to integrate target behavior into the multiple hypothesis tracker (MHT) likelihood ratio. In particular, a periodic track appraisal based on behavior is introduced. The track appraisal uses elementary topological data analysis coupled with basic machine-learning techniques, and it adjusts the traditional kinematic data association likelihood (i.e., track score) using an established formulation for feature-aided data association. The proposed method is tested and demonstrated on synthetic vehicular data representing an urban traffic scene generated by the Simulation of Urban Mobility package. The vehicles in the scene exhibit different driving behaviors. The proposed method distinguishes those behaviors and shows improved data association decisions relative to a conventional, kinematic MHT.},
author = {Bendich, Paul and Chin, Sang Peter and Clark, Jesse and Desena, Jonathan and Harer, John and Munch, Elizabeth and Newman, Andrew and Porter, David and Rouse, David and Strawn, Nate and Watkins, Adam},
doi = {10.1109/TAES.2016.160405},
issn = {1557-9603},
journal = {IEEE Transactions on Aerospace and Electronic Systems},
month = {December},
number = {6},
pages = {2644–2661},
title = {Topological and statistical behavior classifiers for tracking applications},
volume = {52},
year = {2016}
}

Landmark Detection and Tracking in Ultrasound using a CNN-RNN Framework

Cite

@inproceedings{rangamani_landmark_2016,
abstract = {We present a novel framework for landmark detection and tracking in ultrasound
data. Our method employs a convolutional neural network (CNN) encoder-decoder
for landmark detection, coupled with a recurrent neural network (RNN) for encoding information from previous video frames of the object being tracked. We
evaluated our method on the MICCAI CLUST 2015 De Luca u.a. (2015) challenge
dataset, and have achieved promising results.},
author = {Rangamani, Akshay and Xiong, Tao and Nair, Arun and Tran, Trac and Chin, Sang},
month = {December},
title = {Landmark Detection and Tracking in Ultrasound using a CNN-RNN Framework},
year = {2016}
}

Compressive coding via random replicate mirror

Cite

@inproceedings{tran_compressive_2016,
abstract = {We develop a Compressive Sensing (CS) imaging system that uses titled reflective sub-apertures placed at random angles to create replicates of random placement and orientation within the image plane and a variation adopting the beam splitter. We derive efficient methods based on sparse recovery to calibrate the transfer function of the camera from a set of calibrating images, which allows the reducing number of input-output pairs and to reconstruct the scene from random subsampled measurements after calibration. Various experiments are performed to illustrate successful camera calibration and scene reconstruction from sensor output.},
author = {Tran, Dung N. and Liu, Luoluo and Tran, Trac D. and Chin, Sang Peter and Korn, Jeffrey and Hoke, Eric T.},
booktitle = {2016 IEEE Global Conference on Signal and Information Processing (GlobalSIP)},
doi = {10.1109/GlobalSIP.2016.7905821},
month = {December},
pages = {148–152},
title = {Compressive coding via random replicate mirror},
year = {2016}
}

Wi(Deband)-Fi: A Proposal for an Opportunistic Wideband Architecture Based on Wi-Fi

Cite

@inproceedings{noubir_wideband-fi_2016,
abstract = {We propose a system, Wideband-Fi, that is motivated by the information theoretic optimality of impulsive frequency shift keying (I-FSK) in wideband systems, and by the availability of orthogonal frequencies in current Wi-Fi systems. Using orthogonal frequencies to approximate I-FSK, Wideband-Fi is able to use non-contiguous spectrum bands in an agile manner, thus allowing aggressive spectrum scavenging. We present the main principles of our proposal, preliminary demonstration of its feasibility, and discuss challenges, and its connection to cognate problems such as transport-layer use of multiple wireless interfaces, and detection of white spaces.},
address = {New York, NY, USA},
author = {Noubir, Guevara and Medard, Muriel and Chin, Peter},
booktitle = {Proceedings of the 14th ACM International Symposium on Mobility Management and Wireless Access},
doi = {10.1145/2989250.2989256},
isbn = {978-1-4503-4503-3},
keywords = {i-fsk, ofdm, wi-fi, wireless communications},
note = {event-place: Malta, Malta},
pages = {115–122},
publisher = {Association for Computing Machinery},
series = {MobiWac ’16},
title = {Wi(Deband)-Fi: A Proposal for an Opportunistic Wideband Architecture Based on Wi-Fi},
url = {https://doi.org/10.1145/2989250.2989256},
year = {2016}
}

Predicting local field potentials with recurrent neural networks

Cite

@inproceedings{kim_predicting_2016,
abstract = {We present a Recurrent Neural Network using LSTM (Long Short Term Memory) that is capable of modeling and predicting Local Field Potentials. We train and test the network on real data recorded from epilepsy patients. We construct networks that predict multi-channel LFPs for 1, 10, and 100 milliseconds forward in time. Our results show that prediction using LSTM outperforms regression when predicting 10 and 100 millisecond forward in time.},
author = {Kim, Louis and Harer, Jacob and Rangamani, Akshay and Moran, James and Parks, Philip D. and Widge, Alik and Eskandar, Emad and Dougherty, Darin and Chin, Sang Peter},
booktitle = {2016 38th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC)},
doi = {10.1109/EMBC.2016.7590824},
month = {August},
note = {ISSN: 1558-4615},
pages = {808–811},
title = {Predicting local field potentials with recurrent neural networks},
year = {2016}
}

A Fully Integrated Wireless Compressed Sensing Neural Signal Acquisition System for Chronic Recording and Brain Machine Interface

Cite

@article{liu_fully_2016,
abstract = {Reliable, multi-channel neural recording is critical to the neuroscience research and clinical treatment. However, most hardware development of fully integrated, multi-channel wireless neural recorders to-date, is still in the proof-of-concept stage. To be ready for practical use, the trade-offs between performance, power consumption, device size, robustness, and compatibility need to be carefully taken into account. This paper presents an optimized wireless compressed sensing neural signal recording system. The system takes advantages of both custom integrated circuits and universal compatible wireless solutions. The proposed system includes an implantable wireless system-on-chip (SoC) and an external wireless relay. The SoC integrates 16-channel low-noise neural amplifiers, programmable filters and gain stages, a SAR ADC, a real-time compressed sensing module, and a near field wireless power and data transmission link. The external relay integrates a 32 bit low-power microcontroller with Bluetooth 4.0 wireless module, a programming interface, and an inductive charging unit. The SoC achieves high signal recording quality with minimized power consumption, while reducing the risk of infection from through-skin connectors. The external relay maximizes the compatibility and programmability. The proposed compressed sensing module is highly configurable, featuring a SNDR of 9.78 dB with a compression ratio of 8×. The SoC has been fabricated in a 180 nm standard CMOS technology, occupying 2.1 mm × 0.6 mm silicon area. A pre-implantable system has been assembled to demonstrate the proposed paradigm. The developed system has been successfully used for long-term wireless neural recording in freely behaving rhesus monkey.},
author = {Liu, Xilin and Zhang, Milin and Xiong, Tao and Richardson, Andrew G. and Lucas, Timothy H. and Chin, Peter S. and Etienne-Cummings, Ralph and Tran, Trac D. and Van der Spiegel, Jan},
doi = {10.1109/TBCAS.2016.2574362},
issn = {1940-9990},
journal = {IEEE Transactions on Biomedical Circuits and Systems},
month = {August},
number = {4},
pages = {874–883},
title = {A Fully Integrated Wireless Compressed Sensing Neural Signal Acquisition System for Chronic Recording and Brain Machine Interface},
volume = {10},
year = {2016}
}

Compressive high speed flow microscopy with motion contrast (Conference Presentation)

Cite

@inproceedings{bosworth_compressive_2016,
abstract = {High-speed continuous imaging systems are constrained by analog-to-digital conversion, storage, and transmission. However, real video signals of objects such as microscopic cells and particles require only a few percent or less of the full video bandwidth for high fidelity representation by modern compression algorithms. Compressed Sensing (CS) is a recent influential paradigm in signal processing that builds real-time compression into the acquisition step by computing inner products between the signal of interest and known random waveforms and then applying a nonlinear reconstruction algorithm. Here, we extend the continuous high-rate photonically-enabled compressed sensing (CHiRP-CS) framework to acquire motion contrast video of microscopic flowing objects. We employ chirp processing in optical fiber and high-speed electro-optic modulation to produce ultrashort pulses each with a unique pseudorandom binary sequence (PRBS) spectral pattern with 325 features per pulse at the full laser repetition rate (90 MHz). These PRBS-patterned pulses serve as random structured illumination inside a one-dimensional (1D) spatial disperser. By multiplexing the PRBS patterns with a user-defined repetition period, the difference signal y_i=phi_i (x_i – x_i-tau) can be computed optically with balanced detection, where x is the image signal, phi_i is the PRBS pattern, and tau is the repetition period of the patterns. Two-dimensional (2D) image reconstruction via iterative alternating minimization to find the best locally-sparse representation yields an image of the edges in the flow direction, corresponding to the spatial and temporal 1D derivative. This provides both a favorable representation for image segmentation and a sparser representation for many objects that can improve image compression.},
author = {Bosworth, Bryan and Stroud, Jasper R. and Tran, Dung N. and Tran, Trac D. and Chin, Sang and Foster, Mark A.},
booktitle = {High-Speed Biomedical Imaging and Spectroscopy: Toward Big Data Instrumentation and Management},
doi = {10.1117/12.2216602},
editor = {Tsia, Kevin K. and Goda, Keisuke},
keywords = {Microscopy, Pulse shaping, Compressive sensing, Fiber optic information processing, Motion contrast, Structured illumination, Ultrafast imaging, Ultrafast optics},
note = {Backup Publisher: International Society for Optics and Photonics},
pages = {124 — 124},
publisher = {SPIE},
title = {Compressive high speed flow microscopy with motion contrast (Conference Presentation)},
url = {https://doi.org/10.1117/12.2216602},
volume = {9720},
year = {2016}
}

72 MHz A-scan optical coherence tomography using continuous high-rate photonically-enabled compressed sensing (CHiRP-CS)

Cite

@inproceedings{stroud_72_2016,
abstract = {Randomly spectrally patterned laser pulses acquire more information in each sample, allowing for increasing imaging speed independent of detector limitations.},
author = {Stroud, Jasper R. and Bosworth, Bryan T. and Tran, Dung N. and Tran, Trac D. and Chin, Sang and Foster, Mark A.},
booktitle = {Conference on Lasers and Electro-Optics},
doi = {10.1364/CLEO_SI.2016.SM2I.1},
keywords = {Optical coherence tomography, Dispersion, Fourier transforms, High speed photography, High throughput optics, Laser sources, Point spread function, Pulse shaping},
note = {Journal Abbreviation: Conference on Lasers and Electro-Optics},
pages = {SM2I.1},
publisher = {Optica Publishing Group},
title = {72 MHz A-scan optical coherence tomography using continuous high-rate photonically-enabled compressed sensing (CHiRP-CS)},
url = {http://opg.optica.org/abstract.cfm?URI=CLEO_SI-2016-SM2I.1},
year = {2016}
}

Compact all-CMOS spatiotemporal compressive sensing video camera with pixel-wise coded exposure

Publication:

Cite

@article{zhang_compact_2016,
abstract = {We present a low power all-CMOS implementation of temporal compressive sensing with pixel-wise coded exposure. This image sensor can increase video pixel resolution and frame rate simultaneously while reducing data readout speed. Compared to previous architectures, this system modulates pixel exposure at the individual photo-diode electronically without external optical components. Thus, the system provides reduction in size and power compare to previous optics based implementations. The prototype image sensor (127 × 90 pixels) can reconstruct 100 fps videos from coded images sampled at 5 fps. With 20× reduction in readout speed, our CMOS image sensor only consumes 14μW to provide 100 fps videos.},
author = {Zhang, Jie and Xiong, Tao and Tran, Trac and Chin, Sang and Etienne-Cummings, Ralph},
doi = {10.1364/OE.24.009013},
journal = {Opt. Express},
keywords = {Image quality, Imaging systems, Computational imaging, Image sensors, Imaging techniques, Optical components, Sensor performance},
month = {April},
note = {Publisher: Optica Publishing Group},
number = {8},
pages = {9013–9024},
title = {Compact all-CMOS spatiotemporal compressive sensing video camera with pixel-wise coded exposure},
url = {http://opg.optica.org/oe/abstract.cfm?URI=oe-24-8-9013},
volume = {24},
year = {2016}
}

Partial face recognition: A sparse representation-based approach

Cite

@inproceedings{liu_partial_2016,
abstract = {Partial face recognition is a problem that often arises in practical settings and applications. We propose a sparse representation-based algorithm for this problem. Our method firstly trains a dictionary and the classifier parameters in a supervised dictionary learning framework and then aligns the partially observed test image and seeks for the sparse representation with respect to the training data alternatively to obtain its label. We also analyze the performance limit of sparse representation-based classification algorithms on partial observations. Finally, face recognition experiments on the popular AR data-set are conducted to validate the effectiveness of the proposed method.},
author = {Liu, Luoluo and Tran, Trac D. and Chin, Sang Peter},
booktitle = {2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
doi = {10.1109/ICASSP.2016.7472105},
month = {March},
note = {ISSN: 2379-190X},
pages = {2389–2393},
title = {Partial face recognition: A sparse representation-based approach},
year = {2016}
}

Low-rank matrices recovery via entropy function

Cite

@inproceedings{tran_low-rank_2016,
abstract = {The low-rank matrix recovery problem consists of reconstructing an unknown low-rank matrix from a few linear measurements, possibly corrupted by noise. One of the most popular method in low-rank matrix recovery is based on nuclear-norm minimization, which seeks to simultaneously estimate the most significant singular values of the target low-rank matrix by adding a penalizing term on its nuclear norm. In this paper, we introduce a new method that requires substantially fewer measurements needed for exact matrix recovery compared to nuclear norm minimization. The proposed optimization program utilizes a sparsity promoting regularization in the form of the entropy function of the singular values. Numerical experiments on synthetic and real data demonstrates that the proposed method outperforms stage-of-the-art nuclear norm minimization algorithms.},
author = {Tran, Dung N. and Chin, Sang Peter and Tran, Trac D. and Huang, Shuai},
booktitle = {2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
doi = {10.1109/ICASSP.2016.7472441},
month = {March},
note = {ISSN: 2379-190X},
pages = {4064–4068},
title = {Low-rank matrices recovery via entropy function},
year = {2016}
}

Neuromorphic sampling on the SpiNNaker and parallella chip multiprocessors

Cite

@inproceedings{mendat_neuromorphic_2016,
abstract = {We present a bio-inspired, hardware/software architecture to perform Markov Chain Monte Carlo sampling on probabilistic graphical models using energy aware hardware. We have developed algorithms and programming data flows for two recently developed multiprocessor architectures, the SpiNNaker and Parallella. We employ a neurally inspired sampling algorithm that abstracts the functionality of neurons in a biological network and exploits the neural dynamics to implement the sampling process. This algorithm maps nicely on the two hardware systems. Speedups as high as 1000 fold are achieved when performing inference using this approach, compared to algorithms running on traditional engineering workstations.},
author = {Mendat, Daniel R. and Chin, Sang and Furber, Steve and Andreou, Andreas G.},
booktitle = {2016 IEEE 7th Latin American Symposium on Circuits & Systems (LASCAS)},
doi = {10.1109/LASCAS.2016.7451094},
month = {February},
pages = {399–402},
title = {Neuromorphic sampling on the SpiNNaker and parallella chip multiprocessors},
year = {2016}
}

Randomized Minmax Regret for Combinatorial Optimization Under Uncertainty

Cite

@inproceedings{mastin_randomized_2015,
abstract = {The minmax regret problem for combinatorial optimization under uncertainty can be viewed as a zero-sum game played between an optimizing player and an adversary, where the optimizing player selects a solution and the adversary selects costs with the intention of maximizing the regret of the player. The conventional minmax regret model considers only deterministic solutions/strategies, and minmax regret versions of most polynomial solvable problems are NP-hard. In this paper, we consider a randomized model where the optimizing player selects a probability distribution (corresponding to a mixed strategy) over solutions and the adversary selects costs with knowledge of the player’s distribution, but not its realization. We show that under this randomized model, the minmax regret version of any polynomial solvable combinatorial problem becomes polynomial solvable. This holds true for both interval and discrete scenario representations of uncertainty. Using the randomized model, we show new proofs of existing approximation algorithms for the deterministic model based on primal-dual approaches. We also determine integrality gaps of minmax regret formulations, giving tight bounds on the limits of performance gains from randomization. Finally, we prove that minmax regret problems are NP-hard under general convex uncertainty.},
address = {Berlin, Heidelberg},
author = {Mastin, Andrew and Jaillet, Patrick and Chin, Sang},
booktitle = {Algorithms and Computation},
editor = {Elbassioni, Khaled and Makino, Kazuhisa},
isbn = {978-3-662-48971-0},
pages = {491–501},
publisher = {Springer Berlin Heidelberg},
title = {Randomized Minmax Regret for Combinatorial Optimization Under Uncertainty},
year = {2015}
}

Compressed sensing block-wise exposure control algorithm using optical flow estimation

Cite

@inproceedings{xiong_compressed_2015,
abstract = {Recently, CMOS image sensors have attracted more and more attention from the applications of navigation, monitoring and search-and-rescue operations. Specially, CMOS image sensors mounted on insects need to be fast, adaptive to the environment and power efficiency. To simultaneously satisfy both requirements of reconstruction quality and low power consumption, we propose a compressed sensing block-wise exposure control algorithm using optical flow estimation. This framework has been demonstrated to further improve recovery performance (\textgreater 25 dB) with high compression ratio (\textgreater= 10 : 1), which also provides a promising method for real-time CMOS implementation.},
author = {Xiong, Tao and Zhang, Jie and Chin, Sang and Tran, Trac D. and Etienne-Cummings, Ralph},
booktitle = {2015 IEEE Biomedical Circuits and Systems Conference (BioCAS)},
doi = {10.1109/BioCAS.2015.7348443},
month = {October},
pages = {1–4},
title = {Compressed sensing block-wise exposure control algorithm using optical flow estimation},
year = {2015}
}

Local sensing with global recovery

Cite

@inproceedings{tran_local_2015,
abstract = {In this paper, we study Locally Compressed Sensing for images, where sampling process is allowed to be performed on arbitrary local regions of the images. We propose a fast and efficient reconstruction algorithm which utilizes local structures of images. Several numerical experiments on real images demonstrates that our algorithm yields better reconstruction quality than existing techniques at much lower computational complexity and memory requirement.},
author = {Tran, Dung N. and Tran, Duyet N. and Chin, Sang Peter and Tran, Trac D.},
booktitle = {2015 IEEE International Conference on Image Processing (ICIP)},
doi = {10.1109/ICIP.2015.7351620},
month = {September},
pages = {4313–4316},
title = {Local sensing with global recovery},
year = {2015}
}

Neural signal processing and closed-loop control algorithm design for an implanted neural recording and stimulation system

Cite

@inproceedings{hamilton_neural_2015,
abstract = {A fully autonomous intracranial device is built to continually record neural activities in different parts of the brain, process these sampled signals, decode features that correlate to behaviors and neuropsychiatric states, and use these features to deliver brain stimulation in a closed-loop fashion. In this paper, we describe the sampling and stimulation aspects of such a device. We first describe the signal processing algorithms of two unsupervised spike sorting methods. Next, we describe the LFP time-frequency analysis and feature derivation from the two spike sorting methods. Spike sorting includes a novel approach to constructing a dictionary learning algorithm in a Compressed Sensing (CS) framework. We present a joint prediction scheme to determine the class of neural spikes in the dictionary learning framework; and, the second approach is a modified OSort algorithm which is implemented in a distributed system optimized for power efficiency. Furthermore, sorted spikes and time-frequency analysis of LFP signals can be used to generate derived features (including cross-frequency coupling, spike-field coupling). We then show how these derived features can be used in the design and development of novel decode and closed-loop control algorithms that are optimized to apply deep brain stimulation based on a patient’s neuropsychiatric state. For the control algorithm, we define the state vector as representative of a patient’s impulsivity, avoidance, inhibition, etc. Controller parameters are optimized to apply stimulation based on the state vector’s current state as well as its historical values. The overall algorithm and software design for our implantable neural recording and stimulation system uses an innovative, adaptable, and reprogrammable architecture that enables advancement of the state-of-the-art in closed-loop neural control while also meeting the challenges of system power constraints and concurrent development with ongoing scientific research designed to define brain network connectivity and neural network dynamics that vary at the individual patient level and vary over time.},
author = {Hamilton, Lei and McConley, Marc and Angermueller, Kai and Goldberg, David and Corba, Massimiliano and Kim, Louis and Moran, James and Parks, Philip D. and Chin, Sang and Widge, Alik S. and Dougherty, Darin D. and Eskandar, Emad N.},
booktitle = {2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC)},
doi = {10.1109/EMBC.2015.7320207},
month = {August},
note = {ISSN: 1558-4615},
pages = {7831–7836},
title = {Neural signal processing and closed-loop control algorithm design for an implanted neural recording and stimulation system},
year = {2015}
}

Targeted Dot Product Representation for Friend Recommendation in Online Social Networks

Cite

@inproceedings{dao_targeted_2015,
abstract = {In this paper, we develop Targeted Dot Product Representation (TarDPR), a DPR-based feature selection and combination framework for friend recommendation in online social networks (OSNs). Our approach modifies conventional DPR techniques and makes itself applicable to OSNs by focusing on computing a consistent representation while minimizing unnecessary suggestions made outside these interested regions. A notable property of TarDPR is its ability to effectively incorporate different types of social features and produce new meaningful features that help competitive approaches to significantly improve their recommendation quality. We derive an iterative algorithm for TarDPR that is supported by mathematical analysis, and is efficient on large social traces. To certify the usability of our approach, we conduct empirical experiments on real social traces including Facebook and Foursquare social networks. The competitive experimental results show that TarDPR achieves up to 15% improvement in comparison with other competitive methods. These results consequently confirm the efficacy of our suggested framework.},
address = {New York, NY, USA},
author = {Dao, Minh D. and Rangamani, Akshay and Chin, Sang Peter and Nguyen, Nam P. and Tran, Trac D.},
booktitle = {Proceedings of the 2015 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining 2015},
doi = {10.1145/2808797.2809414},
isbn = {978-1-4503-3854-7},
note = {event-place: Paris, France},
pages = {349–356},
publisher = {Association for Computing Machinery},
series = {ASONAM ’15},
title = {Targeted Dot Product Representation for Friend Recommendation in Online Social Networks},
url = {https://doi.org/10.1145/2808797.2809414},
year = {2015}
}

Stochastic Block Model and Community Detection in Sparse Graphs: A spectral algorithm with optimal rate of recovery

Cite

@inproceedings{chin_stochastic_2015,
abstract = {In this paper, we present and analyze a simple and robust spectral algorithm for the stochastic block model with k blocks, for any k fixed. Our algorithm works with graphs having constant edge density, under an optimal condition on the gap between the density inside a block and the density between the blocks. As a co-product, we settle an open question posed by Abbe et. al. concerning censor block models.},
address = {Paris, France},
author = {Chin, Peter and Rao, Anup and Vu, Van},
booktitle = {Proceedings of The 28th Conference on Learning Theory},
editor = {Grünwald, Peter and Hazan, Elad and Kale, Satyen},
month = {July},
pages = {391–423},
publisher = {PMLR},
series = {Proceedings of Machine Learning Research},
title = {Stochastic Block Model and Community Detection in Sparse Graphs: A spectral algorithm with optimal rate of recovery},
url = {https://proceedings.mlr.press/v40/Chin15.html},
volume = {40},
year = {2015}
}

Ultrawideband compressed sensing of arbitrary multi-tone sparse radio frequencies using spectrally encoded ultrafast laser pulses

Publication:

Cite

@article{bosworth_ultrawideband_2015,
abstract = {We demonstrate a photonic system for pseudorandom sampling of multi-tone sparse radio-frequency (RF) signals in an 11.95-GHz bandwidth using <1% of the measurements required for Nyquist sampling. Pseudorandom binary sequence (PRBS) patterns are modulated onto highly chirped laser pulses, encoding the patterns onto the optical spectra. The pulses are partially compressed to increase the effective sampling rate by 2.07×, modulated with the RF signal, and fully compressed yielding optical integration of the PRBS-RF inner product prior to photodetection. This yields a 266× reduction in the required electronic sampling rate. We introduce a joint-sparsity-based matching-pursuit reconstruction via bagging to achieve accurate recovery of tones at arbitrary frequencies relative to the reconstruction basis.},
author = {Bosworth, Bryan T. and Stroud, Jasper R. and Tran, Dung N. and Tran, Trac D. and Chin, Sang and Foster, Mark A.},
doi = {10.1364/OL.40.003045},
journal = {Opt. Lett.},
keywords = {Fiber optics links and subsystems, Ultrafast information processing, Data processing by optical means, Discrete Fourier transforms, Frequency measurement, Microwave photonic filters, Optical processing devices, Phase shift, Ultrafast lasers, Ultrawideband signals},
month = {July},
note = {Publisher: Optica Publishing Group},
number = {13},
pages = {3045–3048},
title = {Ultrawideband compressed sensing of arbitrary multi-tone sparse radio frequencies using spectrally encoded ultrafast laser pulses},
url = {http://opg.optica.org/ol/abstract.cfm?URI=ol-40-13-3045},
volume = {40},
year = {2015}
}

Ultrawideband RF compressed sensing using spectrally-encoded ultrafast laser pulses

Cite

@inproceedings{bosworth_ultrawideband_2015-1,
abstract = {We experimentally demonstrate a photonic RF sampling system that utilizes chirp processing of ultrafast laser pulses to achieve all-optical high-rate pseudorandom patterning and inner product integration for compressed sensing measurement. We successfully acquire multi-tone sparse radio frequency (RF) signals at arbitrary offsets from the reconstruction basis frequencies in an 11.95 GHz bandwidth utilizing less than 1% of the measurements traditionally required for Nyquist sampling. Pseudorandom binary sequence (PRBS) patterns are modulated onto time-stretched optical pulses, encoding them onto the optical spectra at a rate of one unique pattern per pulse. Thus patterned, the pulses are then partially-compressed, increasing the system’s effective sampling rate by 2.07×, well beyond the electronic modulation rate. The partially-compressed patterned pulses are then modulated again with the RF signal under test and fully compressed to perform optical integration of the PRBS-RF inner product before output photodetection and digitization. This achieves a reduction in required electronic sampling rate by two orders of magnitude.},
author = {Bosworth, Bryan T. and Stroud, Jasper R. and Tran, Dung N. and Tran, Trac D. and Chin, Sang and Foster, Mark A.},
booktitle = {2015 3rd International Workshop on Compressed Sensing Theory and its Applications to Radar, Sonar and Remote Sensing (CoSeRa)},
doi = {10.1109/CoSeRa.2015.7330269},
month = {June},
pages = {85–88},
title = {Ultrawideband RF compressed sensing using spectrally-encoded ultrafast laser pulses},
year = {2015}
}