keyulu

Keyulu Xu

Email: keyulu [at] mit (dot) edu

Office: MIT Stata Center, 32-G480

Mail: 2-1-2 Hitotsubashi, Chiyoda-ku, Tokyo 101-8430


I am a researcher, trader, and entrepreneur. My current goal is to revolutionize the global financial system. I received my Ph.D. in EECS from MIT, where I was affiliated with CSAIL and the Machine Learning group.

At MIT, I was advised by Stefanie Jegelka. Previously, I was an undergraduate at UBC, advised by Nick Harvey. I seasonally visited Ken-ichi Kawarabayashi at NII in Tokyo. I will be appointed as a visiting professor (not for business or academic purposes). Please feel free to contact me for collaboration or open intern positions.

Recent News

Publications

Email me if you have any questions about my papers or code, or if you would like to collaborate with me.

  1. Modeling Intelligence via Graph Neural Networks
    MIT Ph.D. Thesis, 2021.
    [Paper (soon)] [Slides]
  2. Optimization of Graph Neural Networks: Implicit Acceleration by Skip Connections and More Depth
    Keyulu Xu, Mozhi Zhang, Stefanie Jegelka, Kenji Kawaguchi.
    International Conference on Machine Learning (ICML) 2021.
    [Paper] [arXiv]
  3. Information Obfuscation of Graph Neural Networks
    Alex Liao, Han Zhao, Keyulu Xu, Tommi Jaakkola, Geoffrey Gordon, Stefanie Jegelka, Russ Salakhutdinov.
    International Conference on Machine Learning (ICML) 2021.
    [Paper] [arXiv] [code]
  4. GraphNorm: A Principled Approach to Accelerating Graph Neural Network Training
    Tianle Cai, Shengjie Luo, Keyulu Xu, Di He, Tie-Yan Liu, Liwei Wang.
    International Conference on Machine Learning (ICML) 2021.
    [Paper] [arXiv] [code]
  5. Noisy Labels Can Induce Good Representations
    Jingling Li, Mozhi Zhang, Keyulu Xu, John P. Dickerson, Jimmy Ba.
    Manuscript 2021.
    [Paper] [arXiv] [code]
  6. How Neural Networks Extrapolate: From Feedforward to Graph Neural Networks
    Keyulu Xu, Mozhi Zhang, Jingling Li, Simon S. Du, Ken-ichi Kawarabayashi, Stefanie Jegelka.
    International Conference on Learning Representations (ICLR) 2021 (Oral).
    [Paper] [arXiv] [code]
  7. What Can Neural Networks Reason About?
    Keyulu Xu, Jingling Li, Mozhi Zhang, Simon S. Du, Ken-ichi Kawarabayashi, Stefanie Jegelka.
    International Conference on Learning Representations (ICLR) 2020 (Spotlight).
    [Paper] [arXiv] [code]
  8. Graph Neural Tangent Kernel: Fusing Graph Neural Networks with Graph Kernels
    Simon S. Du, Kangcheng Hou, Barnabas Poczos, Ruslan Salakhutdinov, Ruosong Wang, Keyulu Xu.
    Advances in Neural Information Processing Systems (NeurIPS) 2019.
    [Paper] [arXiv] [code]
  9. Are Girls Neko or Shōjo?
    Cross-Lingual Alignment of Non-Isomorphic Embeddings with Iterative Normalization

    Mozhi Zhang, Keyulu Xu, Ken-ichi Kawarabayashi, Stefanie Jegelka, Jordan Boyd-Graber.
    Association for Computational Linguistics (ACL) 2019.
    [Paper] [arXiv] [code]
  10. How Powerful are Graph Neural Networks?
    Keyulu Xu, Weihua Hu, Jure Leskovec, Stefanie Jegelka.
    International Conference on Learning Representations (ICLR) 2019 (Oral).
    [Paper] [arXiv] [code]
  11. Representation Learning on Graphs with Jumping Knowledge Networks
    Keyulu Xu, Chengtao Li, Yonglong Tian, Tomohiro Sonobe, Ken-ichi Kawarabayashi, Stefanie Jegelka.
    International Conference on Machine Learning (ICML) 2018 (Long Talk).
    [Paper] [arXiv] [code]
  12. Distributional Adversarial Networks
    Chengtao Li, David Alvarez-Melis, Keyulu Xu, Stefanie Jegelka, Suvrit Sra.
    International Conference on Learning Representations workshop track (ICLR) 2018.
    [arXiv] [code]
  13. Generating Random Spanning Trees via Fast Matrix Multiplication
    Nicholas J. A. Harvey and Keyulu Xu.
    Latin American Theoretical Informatics Symposium (LATIN) 2016.
    [Paper] [Web]

Experience

My academic and business experience mostly happen in NYC and Tokyo.

Talks