Shenghui Chen

Shenghui (Vivian) Chen

I am a final-year Ph.D. student in Computer Science at the University of Texas at Austin, advised by Ufuk Topcu. Previously, I earned my B.S. in Computer Science and Mathematics at the University of Virginia, advised by Lu Feng.

I develop AI agents that collaborate with people under incomplete information and human cognitive limits. As agents become more capable, people still provide incomplete guidance and have limited attention to review their decisions. My research addresses both sides of this exchange: how agents infer human intent, and what information they should communicate to support human decisions.

My research spans the following topics:

Human-Centered Evaluation and Information Selection (3)
  • Overview of VEGAS: Human-Aligned Video Caption Evaluation via Gaze
    VEGAS: Human-Aligned Video Caption Evaluation via Gaze

    Shenghui Chen, Po-han Li, Ximeng Sun, Shijia Yang, Emad Barsoum, Zicheng Liu, Sandeep Chinchali, Ufuk Topcu

    Transactions on Machine Learning Research (TMLR), 2026 · (Accepted for publication)

    paper bibtex
  • Overview of VIBE: Annotation-Free Video-to-Text Information Bottleneck Evaluation for TL;DR
    VIBE: Annotation-Free Video-to-Text Information Bottleneck Evaluation for TL;DR

    Shenghui Chen*, Po-han Li*, Sandeep Chinchali, Ufuk Topcu

    Conference on Neural Information Processing Systems (NeurIPS), 2025

    paper bibtex code project
  • Overview of What We are Missing in Multimodal LLM Evaluation?
    What We are Missing in Multimodal LLM Evaluation?

    Po-han Li*, Shenghui Chen*, Sandeep Chinchali, Ufuk Topcu

    Communications of the ACM, 2026 · (Accepted for publication)

    paper bibtex
Planning with Human Intent under Incomplete Information (3)
  • Overview of Human-Agent Cooperation in Games under Incomplete Information through Natural Language Communication
    Human-Agent Cooperation in Games under Incomplete Information through Natural Language Communication

    Shenghui Chen, Daniel Fried, Ufuk Topcu

    International Joint Conference on Artificial Intelligence (IJCAI), Human-Centred AI, 2024

    paper doi bibtex code
  • Overview of Human-Agent Coordination in Games under Incomplete Information via Multi-Step Intent
    Human-Agent Coordination in Games under Incomplete Information via Multi-Step Intent

    Shenghui Chen*, Ruihan Zhao*, Sandeep Chinchali, Ufuk Topcu

    International Conference on Autonomous Agents and Multiagent Systems (AAMAS), 2025

    paper doi bibtex code project
  • Overview of IG-MCTS: Human-in-the-Loop Cooperative Navigation under Incomplete Information
    IG-MCTS: Human-in-the-Loop Cooperative Navigation under Incomplete Information

    Shenghui Chen*, Ruihan Zhao*, Sandeep Chinchali, Ufuk Topcu

    International Conference on Autonomous Agents and Multiagent Systems (AAMAS), 2026 · (Extended Abstract)

    paper bibtex
Inverse Game and Mechanism Design for Multi-Agent (3)
  • Overview of Cost Design in Atomic Routing Games
    Cost Design in Atomic Routing Games

    Yue Yu, Shenghui Chen, David Fridovich-Keil, Ufuk Topcu

    American Control Conference (ACC), 2023

    paper doi bibtex code blog
  • Overview of Relationship Design for Socially-Aware Behavior in Static Games
    Relationship Design for Socially-Aware Behavior in Static Games

    Shenghui Chen*, Yigit E. Bayiz*, David Fridovich-Keil, Ufuk Topcu

    Journal of Autonomous Agents and Multi-Agent Systems (JAAMAS), 2025

    paper bibtex code
  • Overview of Soft-Bellman Equilibrium in Affine Markov Games: Forward Solutions and Inverse Learning
    Soft-Bellman Equilibrium in Affine Markov Games: Forward Solutions and Inverse Learning

    Shenghui Chen, Yue Yu, David Fridovich-Keil, Ufuk Topcu

    Conference on Decision and Control (CDC), 2023

    paper doi bibtex code

* Equal contribution.

See the complete list of publications.

Open-Source Datasets

VEGAS 2026

Egocentric activities and instructional slides paired with synchronized gaze and reference annotations for attention-aligned video caption evaluation.

Public download currently unavailable.

Explore the datasets.

News

Older updates

  • Invited talk to Prof. Mykel Kochenderfer’s group, SISL, at Stanford University.

  • Invited talk to Prof. Amy Pavel’s group at UC Berkeley.

  • Invited talk at Prof. Changliu Liu’s Intelligent Control Lab at Carnegie Mellon University.

  • Wrapped up a rewarding internship with AMD’s GenAI group.

  • Glad to be shortlisted for the ADF 2025 Rising Stars Asia program, see here.

  • Our paper on annotation-free video-to-text evaluation has been accepted to NeurIPS 2025! Learn more on our project page.

  • Check out our new preprint on annotation-free video-to-text evaluation, introducing VIBE to select captions as human-centered TL;DRs

  • Presented at the AIML Seminar, University of Virginia on title “Human-Agent Collaboration under Incomplete Information: Intent, Communication, and Planning”

  • Presented at the NSF CPS Frontier Site Visit, Purdue University on title “Human-Agent Coordination in Games under Incomplete Information via Multi-Step Intent”

Outreach Activities

Selected Honors

  • ICML Silver Reviewer Award — International Conference on Machine Learning
  • Rising Stars Women in Engineering — Asian Deans’ Forum · Shortlisted
  • Outstanding Undergraduate Researcher — Computing Research Association · Honorable Mention
  • Outstanding Winner & INFORMS Award — Interdisciplinary Contest in Modeling · Top 0.5%