Profile Picture

Shiyun (Joy) Cheng

B.S. in Computer Science and Data Science
University of Wisconsin, Madison

Hi, I’m Shiyun Cheng, also known as Joy. I’m currently advancing my research in AI at the University of Wisconsin-Madison under the mentorship of Prof. Jerry Zhu, Prof. Kirthevasan Kandasamy, and Dr. Young Wu. My interests broadly encompass artificial intelligence, game theory, and computational optimization. I’m particularly focused on understanding the strengths and limitations of large language models and exploring how optimization and algorithms can be designed to make AI more adaptable and effective in complex environments.

Research

  • Integer Optimization for Achieving Unique Nash Equilibria
  • Algorithm for Dynamic Pricing with Uncertain Buyer Distribution
  • A Visual Theory-of-Mind Benchmark for Multimodal Large Language Models
  • Deep Q-Network (DQN) Analysis and Application

    • Teaching

  • Undergraduate TA for Class of Intro to Algorithms (CS577) in University of Wisconsin-Madison
  • Undergraduate TA for Class of Data Programming II (CS320) in University of Wisconsin-Madison

    • Publications

    2024
    Publication Thumbnail
    CHARTOM: A Visual Theory-of-Mind Benchmark for Multimodal Large Language Models

    Shubham Bharti*, Shiyun Cheng*, Jihyun Rho, Martina Rao, Xiaojin Zhu

    We introduce CHARTOM, a visual theory-of-mind benchmark for multimodal large language models. CHARTOM consists of specially designed data visualizing charts. Given a chart, a language model needs to not only correctly comprehend the chart (the FACT question) but also judge if the chart will be misleading to a human reader (the MIND question). Both questions have significant societal benefits. We detail the construction of the CHARTOM benchmark including its calibration on human performance.

    Paper

    Contact

    You can contact me via email at scheng227@wisc.edu.