Publications

• 2024

• Analyzing Large Language Model Behavior via Embedding Analysis BIB
  Sourya Dey, Michael Robinson, Shauna Sweet, Andrew Lauziere, Jonathan Daugherty, Caitlin Burgess
  To appear in Applied Human Factors and Ergonomics: International Conference on Human Factors in Design, Engineering, and Computing (AHFE Hawaii Edition)
• The structure of the token space for large language models BIB
  Michael Robinson, Sourya Dey, Shauna Sweet
  arXiv Differential Geometry; arXiv Artificial Intelligence

• 2023

• The 2010 Census Confidentiality Protections Failed, Here's How and Why BIB arXiv Github
  John M. Abowd, Tamara Adams, Robert Ashmead, David Darais, Sourya Dey, Simson L. Garfinkel, Nathan Goldschlag, Daniel Kifer, Philip Leclerc, Ethan Lew, Scott Moore, Rolando A. Rodríguez, Ramy N. Tadros, Lars Vilhuber
  National Bureau of Economic Research (NBER)
• DLKoopman: A deep learning software package for Koopman theory BIB arXiv PyPI Github Poster
  Sourya Dey, Eric Davis
  5th Annual Learning for Dynamics & Control (L4DC) Conference
• Leveraging Manifold Learning and Relationship Equity Management for Symbiotic Explainable Artificial Intelligence BIB Slides
  Sourya Dey, Adam Karvonen, Ethan Lew, Donya Quick, Panchapakesan Shyamshankar, Ted Hille, Matt LeBeau, Eric Davis
  9th International Conference on Human Factors in Robots, Drones and Unmanned Systems -- affiliated to 14th International Conference on Applied Human Factors and Ergonomics (AHFE)

• 2022

• LAGOON: An Analysis Tool for Open Source Communities BIB arXiv Github Slides Video
  Sourya Dey, Walt Woods
  Mining Software Repositories (MSR) Conference
• Dynamic Mode Decomposition and Koopman Theory BIB
  Sourya Dey
  arXiv Optimization and Control

• 2021

• Deep-n-Cheap: An Automated Efficient and Extensible Search Framework for Cost-Effective Deep Learning BIB Github
  Sourya Dey, Sara Babakniya, Saikrishna C. Kanala, Marco Paolieri, Leana Golubchik, Peter A. Beerel, Keith M. Chugg
  Springer Nature (SN) Computer Science journal

• 2020

• Deep-n-Cheap: An Automated Search Framework for Low Complexity Deep Learning BIB arXiv Github Video
  Sourya Dey, Saikrishna C. Kanala, Keith M. Chugg, Peter A. Beerel
  Asian Conference on Machine Learning (ACML)

• 2019

• Pre-Defined Sparse Neural Networks with Hardware Acceleration BIB arXiv Github
  Sourya Dey, Kuan-Wen Huang, Peter A. Beerel, Keith M. Chugg
  IEEE Journal on Emerging and Selected Topics in Circuits and Systems (JETCAS)

• 2018

• Morse Code Datasets for Machine Learning BIB arXiv Github Slides
  Sourya Dey, Keith M. Chugg, Peter A. Beerel
  Won Best Paper Award at International Conference on Computing, Communication and Networking Technologies (ICCCNT)
  Dataset is also available on IEEEDataPort and has its own data competition!
• Characterizing Sparse Connectivity Patterns in Neural Networks BIB arXiv
  Sourya Dey, Kuan-Wen Huang, Peter A. Beerel, Keith M. Chugg
  Information Theory and Applications (ITA) Workshop
• A Highly Parallel FPGA Implementation of Sparse Neural Network Training BIB arXiv Github Poster
  Sourya Dey, Diandian Chen, Zongyang Li, Souvik Kundu, Kuan-Wen Huang, Keith M. Chugg, Peter A. Beerel
  International Conference on Reconfigurable Computing and FPGAs (ReConFig)
  See arXiv for expanded pre-print version

• 2017

• Accelerating Training of Deep Neural Networks via Sparse Edge Processing BIB arXiv Slides
  Sourya Dey, Yinan Shao, Keith M. Chugg, Peter A. Beerel
  International Conference on Artificial Neural Networks (ICANN)
  First paper on the concept of pre-defined sparsity
• Interleaver Design for Deep Neural Networks BIB arXiv Slides
  Sourya Dey, Peter A. Beerel, Keith M. Chugg
  Asilomar Conference on Signals, Systems and Computers
• Pricing Football Players using Neural Networks BIB Github Slides
  Sourya Dey
  arXiv Machine Learning

• 2016

• A Nonuniform Sampling ADC Architecture With Reconfigurable Digital Anti-Aliasing Filter BIB
  Tzu-Fan Wu, Sourya Dey, Mike Shuo-Wei Chen
  IEEE Transactions on Circuits and Systems (TCAS) I : Regular Papers

---

Invited Talks, Posters, and other Media

• DLKoopman: A deep learning software package for Koopman theory
  • Workshop on Symbiotic AI for Robust Digital Engineering, May 2023
• ORCA: Orchestrating Symbiotic Intelligence for Agile and Adaptable, Crisis Response Decision Making
  • Chesapeake Large-Scale Analytics Conference (CLSAC), Oct 2022
• Open is not always welcoming: Examining how toxic interpersonal signals impact and reflect collaboration and prestige in the open source Python language development community
  • International Network for Social Network Analysis (INSNA) Sunbelt Conference, Jul 2022
• Exploring Complexity Reduction to Address AI's Carbon Footprint
  • Galois Quotidian Machine Ethics (QME) workshop, Feb 2022 Slides
• Deep-n-Cheap: An Automated Search Framework for Low Complexity Deep Learning
  • iRobot Webinar on Robot Intelligence Technology and Applications, Jun 2020 Slides
  • Samsung Research America, May 2020 Slides
  • Galois, Apr 2020 Slides
  • Deep Learning course (EE599) at USC, hosted by Prof. Keith Chugg, Spring 2020 Slides
• Exploring Complexity Reduction in Deep Learning
  • Rice University, hosted by Prof. Genevera Allen, ECE Seminar Series, Feb 2020 Slides
  • Indian Institute of Technology Kharagpur, hosted by Prof. Debdoot Sheet, IEEE Student Branch and Center for Excellence in Artificial Intelligence, Jan 2020 Slides
  • USC Information Sciences Institute (ISI) Research Day, Jan 2020 Poster
  • Arizona State University, hosted by Prof. Jae-sun Seo, Nov 2019 Slides
  • USC Ming Hsieh ECE Department Annual Research Festival, Nov 2019 Poster
• Hardware Acceleration of Neural Networks via Pre-defined Sparsity
  • Hardware and Algorithms for Learning On-a-Chip (HALO) workshop, International Conference on Computer Aided Design (ICCAD), 2017 Poster
• Reduced Complexity Machine Learning
  • Mining Software Repositories (MSR) conference, 2022 Slides

---

Dissertation Material

• Exploring Complexity Reduction in Deep Learning Slides
  PhD Dissertation (Defense)
  Committee Members: Peter Beerel, Keith Chugg, Leana Golubchik
  April 22nd, 2020
• Exploring Complexity Reduction for Learning in Deep Neural Networks Slides
  PhD Dissertation Proposal (Qualifying Exam)
  Committee Members: Peter Beerel, Keith Chugg, Leana Golubchik, Panayiotis Georgiou, Mahdi Soltanolkotabi
  April 23rd, 2019

---

Reports and Additional Projects

• Semantic Segmentation of Ultra-Resolution 3D Microscopic Neural Imagery Github Slides
  Hsiao-Lun Wang, Muye Zhu, Sourya Dey, Ting-Ru Lin
  Course Project for Deep Learning and its Applications, USC, 2017
• Question-Answer User Recommender System Github
  David Beccera, Elizabeth Boroson, Sourya Dey
  Submission for ByteCup Challenge and Course Project for Machine Learning, USC, 2016
• Programming Language Speed Tests Github
  Sourya Dey
  Repository to test and compare speed and efficiency of different programming language methods and frameworks, 2019 –

---

Successful Grant Funding Proposals

(assisted in writing)

• Training Sparse Neural Networks with Co-Designed Hardware Accelerators: Enabling Model Optimization and Scientific Exploration
  K. M. Chugg, P. A. Beerel, L. Golubchik, P. A. Georgiou
  National Science Foundation Core Programs, 2017
• Scalable Acceleration Platform Integrating Reconfigurable Computing and Natural Language Processing Technologies (SAPIENT)
  A. Schmidt, P. A. Beerel
  Defense Threat Reduction Agency, 2019