Chundi Liu

I am a machine learning scientist at Layer6 AI, where I work on computer vision, biomedical image analysis and machine learning applications in capital market.

I obtained my master's degree from University of Toronto in 2018 and honors bachelor's degree from China Agricultural University in 2015.

We are hiring Machine Learning Scientists and Data Engineers. Feel free to reach out to me or directly apply from here.

Email  /  Google Scholar  /  Kaggle  /  Github  /  LinkedIn

profile photo

I have always been passionate about machine learning competitions. Currently, I am a Kaggle master with best rank 70/125,000. Here are some highlights:

  • 4th place of WIDER Face & Person Challenge 2019 - Track 3: Cast Search by Portrait. [poster]
  • 3rd place of Google Landmark Retrieval 2019. [paper]
  • 4th place (out of 1500) of RSNA Pneumonia Detection Challenge.
  • 2nd place of Google Landmark Retrieval Challenge. [paper][poster]

  • Research

    My research interests broadly lie in machine learning, computer vision and natural language processing, with a recent focus on image retrieval, graph modelling and unsupervised learning. I'm also into machine learning applications in biomedical area.

    Guided Similarity Separation for Image Retrieval
    Chundi Liu, Guangwei Yu, Cheng Chang, Himanshu Rai, Junwei Ma,
    Satya Krishna Gorti, Maksims Volkovs
    Neural Information Processing Systems (NeurIPS), 2019 (Oral 2.5%)

    A new unsupervised model for image retrieval based on graph neural network neighbor encoding and a novel guided similarity separation loss. Results on public benchmarks show highly competitive performance with up to 25% relative improvement.

    Explore-Exploit Graph Traversal for Image Retrieval
    Cheng Chang*, Guangwei Yu*, Chundi Liu, Maksims Volkovs
    Computer Vision and Pattern Recognition (CVPR), 2019

    A new approach for image retrieval based on graph traversal by alternating between explore and exploit steps to capture the data manifold.

    Unsupervised Document Embedding With CNNs
    Chundi Liu*, Shunan Zhao*, Maksims Volkovs
    arXiv, 2017

    A CNN model for learning document embedding in an unsupervised fashion.

    Oral Presentation, NeurIPS 2019

    Vancouver, BC, Canada

    Research and Careers in AI, Vector Institute

    Toronto, ON, Canada

    Press Coverage
    U of T alumni and graduate students part of Layer 6 AI's win in global competition

    U of T news

    Website template comes from Jon Barron.