We have two goals:

  • Machine learning for science: Develop machine learning methods that help scientists finding new scientific discoveries.
  • Machine learning: Develop fundamental machine learning algorithms.

Sparse Learning

  • High-dimensional nonlinear feature selection
  • Matrix completion

Selective Inference for Kernels

  • Selective Independence test
  • Selective two-sample test
  • Selective goodness of fit test

Optimal Transport

  • OT for tree-structured data
  • OT for high-dimensional data
  • OT based Semantic correspondence