Hanming Ye

I’m building Waddle Labs with Yiding Song. We’re an applied AI lab building LLMs that control robots.

I studied math at Harvard, where I was a TA for Math 55. In research, I’m interested in what we can learn about the brain by observing and building machine learning models.

I’m grateful to have worked on algebraic topology with Dr. David J. Lee at RSI/MIT, predictive coding with Dr. Chenguang Li, spiking neural networks with Jorin Overwiening and Prof. Haim Sompolinsky, and robot learning with Prof. Yilun Du. I’m also thankful for Quincy and my roommates. Much of this work happened there.

You can reach me via .

Publications

  1. 2026
    Few-shot Task Learning via Compositional Concept Inference

    Hanming Ye, Yiding Song, Yilun Du

    NeurIPS 2026

  2. 2026
    Model Capacity Determines Grokking through Competing Memorisation and Generalisation Speeds

    Yiding Song, Hanming Ye

    NeurIPS 2026 paper

  3. 2026
    Value-driven behaviors emerge from local learning rules in minimal neural networks using online interactions

    Chenguang Li, Hanming Ye, Joanna Aloor, Chunyu A. Duan, Jeffrey Erlich

    COSYNE 2026

  4. 2026
    Tempotron learning in deep neural networks with direct feedback alignment

    Jorin Overwiening, Hanming Ye, M Ganesh Kumar, Haim Sompolinsky

    COSYNE 2026

  5. 2025
    Steering Diffusion Policies with Value-Guided Denoising

    Hanming Ye

    NeurIPS 2025, Workshop on Embodied World Models for Decision Making paper code

  6. 2023
    The Stable Picard Groups of the Exterior Algebras E(n)

    Hanming Ye, David J. Lee

    MIT Research Science Institute paper

    RSI Top 5 Paper Prize