Welcome to the DeLTA Lab

Research Overview

1. We integrate AI methods with physics knowledge.

a) We build LLMs for autoformalization, program synthesis, and reasoning on PDE problems [1, 2].

b) We introduce physics into data-driven fluids modeling and reconstruction [3, 4, 5, 6].

2. We develop principled deep learning methods [7, 8, 9, 10] inspired by deep learning theory.

3. We are broadly interested in LLM [11, 12, 13], computer vision [14, 15], model efficiency [16, 17], self-supervised learning [18, 19], automated machine learning (AutoML) [20], etc.

News

[12/2025]

• Our ICLR 2026 workshop proposal, “Foundation Models for Science,” is accepted. Please consider submitting your great work to our workshop (OpenReview)!

• Our CVPR 2026 workshop proposal, “3D Geometry Generation for Scientific Computing,” is accepted. Please consider submitting your great work to our workshop (OpenReview)!

[10-12/2025] Talks at Oden Institute @ UT Austin, CAIDA @ UBC (Video), EECS 542 @ UMich, PurPL @ Purdue, AI-Math @ Rutgers

[9/2025] Dr. Chen is honored to receive the 2025 Nvidia Academic Grant Award on Physics-Informed Machine Learning.

[8/2025] Dr. Chen will serve as an area chair at ICLR 2026.

[5/2025] PDE-Controller is accepted in ICML’2025

[4/2025] Dr. Chen will serve as an area chair at NeurIPS 2025.

[3/2025] Talks at AAAI 2025, SIAM CSE25, BLISS seminar (Berkeley), TPC Seminar (Argonne National Lab)

[2/2025]

• PDE-Controller is on arXiv now! This is the first AI-for-math work that enables LLM-based scientific reasoning. More data/code/model releases are coming soon!

• Talks at Georgia Tech, NYU

[1/2025] 1 paper accepted in ICLR’2025

[12/2025]

• Dr. Chen is selcted by the AAAI-2025 New Faculty Highlights program.

• Our “Foundation Models for Science” workshop @ NeurIPS 2024 was a success! See video recordings and accepted papers.

[11/2024]

• Dr. Chen will serve as an area chair at ICML 2025.

• Invited talk at 2024 Canadian Mathematical Society (CMS) Winter Meeting

• Invited talk at SFU Mathematics of Computation, Application and Data (“MOCAD”) Seminar

[10/2024] Invited talk at Challenges & opportunities in Foundation Models: accelerating scientific discovery, explainability, efficiency and security, University of Geneva

[9/2024] 1 paper accepted in NeurIPS’2024

[7/2024] Our NeurIPS 2024 workshop proposal, “Foundation Models for Science: Progress, Opportunities, and Challenges,” is accepted. Please consider submitting your great work to our workshop (OpenReview)!

[6/2024]

• Dr. Chen is humbled and grateful that his dissertation, “Demystifying deep network architectures: from theory to applications,” received the INNS 2024 Doctoral Dissertation Award and the iSchools Doctoral Dissertation Award 2024. See also the UT ECE news.
  

• Invited talk at 3rd Workshop on Federated Learning for Computer Vision in CVPR 2024.

• Invited panel session at Learning Deep Low-Dimensional Models from High-Dimensional Data: From Theory to Practice in CVPR 2024.

[5/2024] Dr. Chen is grateful to receive the Google GCP Credit Award (Gemma Academic Program)

[4/2024]

• The DeLTA group welcomes 3 new members: Yuqiu Liu, Jialin Song, Mauricio Soroco

• Invited talk at AI4Science Seminar Series, Amazon AWS

[3/2024]

• Dr. Chen is grateful to receive the NSERC Discovery Grant
  

• 1 paper accepted in ICLR workshop AI4DiffEqtnsInSci

[2/2024] 1 paper accepted in CVPR’2024

[1/2024] 2 papers accepted in ICLR’2024