publications

* denotes equal contribution.

2026

  1. Universality, Function Composition, and Algorithm Emulation All In-Context
    Hong-Yu Chen*, Po-Chiao Lin*, Maojiang Su, Jerry Yao-Chieh Hu, and Han Liu
    In ICML, 2026
  2. Universal Approximation with Softmax Attention
    Jerry Yao-Chieh Hu*, Hude Liu*, Hong-Yu Chen*, Weimin Wu, and Han Liu
    In ICML, 2026
    arXiv
  3. Chain-of-Thought Gradient Descent
    Hong-Yu Chen*, Venkat Sripad Ganti*, Jerry Yao-Chieh Hu, Hude Liu, and Han Liu
    In ICML, 2026
  4. StarEmbed: Benchmarking Time Series Foundation Models on Astronomical Observations of Variable Stars
    Weijian Li*, Hong-Yu Chen*, Nabeel Rehemtulla*, Ved G. Shah, Dennis Wu, Dongho Kim, Qinjie Lin, Adam A. Miller, and Han Liu
    In ICML, 2026
    arXiv Code Website

2025

  1. Transformers versus the EM Algorithm in Multi-class Clustering
    Yihan He, Hong-Yu Chen, Yuan Cao, Jianqing Fan, and Han Liu
    arXiv preprint arXiv:2502.06007, 2025
    arXiv
  2. Learning spectral methods by transformers
    Yihan He, Yuan Cao, Hong-Yu Chen, Dennis Wu, Jianqing Fan, and Han Liu
    arXiv preprint arXiv:2501.01312, 2025
    arXiv
  3. Transformers Simulate MLE for Sequence Generation in Bayesian Networks
    Yuan Cao, Yihan He, Dennis Wu, Hong-Yu Chen, Jianqing Fan, and Han Liu
    arXiv preprint arXiv:2501.02547, 2025
    arXiv

2024

  1. Outlier-Efficient Hopfield Layers for Large Transformer-Based Models
    Jerry Yao-Chieh Hu, Pei-Hsuan Chang, Haozheng Luo, Hong-Yu Chen, Weijian Li, Wei-Po Wang, and Han Liu
    In ICML, 2024
    arXiv PDF Code Poster Slides