Agam Goyal

Apr 2024: Named an exceptional reviewer by DMLR!
Apr 2024: I will be starting as a Ph.D. student at the University of Illinois Urbana-Champaign (UIUC) in Fall 2024.
Apr 2024: Our work on Wisdom of Partisan Crowds of LLM Agents has been accepted to CogSci 2024!
Apr 2024: Our work on Uncovering the Hidden Cost of Model Compression has been accepted to Prompting in Vision Workshop at CVPR 2024!
Mar 2024: Our work on Simulating Opinion Dynamics using LLM Agents has been accepted to NAACL 2024 (Findings).
Mar 2024: Two works on social simulations using LLMs accepted to the LLM Agents workshop at ICLR 2024.
Feb 2024: Serving as a reviewer for ICLR 2024 workshops on LLM Agents and Data-centric Machine Learning.
Nov 2023: Our preprint on a latent linear model for nonlinear coupled oscillators on graphs is out on arXiv.
Nov 2023: Honored to receive the Trewartha Senior Honors Thesis Research Award and the Bromley Conference Travel Award from the L&S Honors committee.
Nov 2023: Two new preprints on using generative agents to study Opinion Dynamics and Wisdom of Partisan Crowds are now out on arXiv!
Nov 2023: Gave a lightning talk on LLM Group Dynamics: A New Approach to Social Simulations at the 2023 Wisconsin Institute for Discovery Symposium. [ Slides ]
Oct 2023: Gave a flash talk on Reprogramming Under Constraints at ML Collective Research Jam. [ Slides ] [ Talk ]
Oct 2023: Gave a talk on Reprogramming Under Constraints at Google Sparsity Reading Group. [ Slides ]
Sep 2023: Our work on an NCPD-based framework to capture spatiotemporal patterns in coupled oscillator and opinion dynamics has been accepted to CNA 2023!
Aug 2023: Check out our pre-print Reprogramming Under Constraints on arXiv!
May 2023: Starting a new position as an SDE Intern at Amazon.
May 2023: Honored to receive the Hilldale Undergraduate/Faculty Research Fellowship, 2023-24 from the Undergraduate Academic Awards Department.

My research interests range across various domains under machine learning and deep learning, including development and utilization of Matrix and Tensor Decomposition techniques to understand emergent or latent properties of complex systems, Network Compression (model sparsity), Computer Vision (medical imaging and model reprogramming), and Natural Language Processing (cognitive sciences, simulating generative agent worlds). I am fortunate to be advised by Dr. Hanbaek Lyu and Dr. Junjie Hu.

Updates

Research

Publications

*indicates equal contribution

I. Towards Constructing Human-like LLM Agents

II. Interpretable Prediction of Outcomes of Dynamical Systems

III. Visual Prompting

Professional Experience

Education