Tam LE  [Ph.D. 2016]

Tam Le Assistant Professor Department of Advanced Data Science Institute of Statistical Mathematics (ISM) Email: tam[at]ism.ac.jp CV - Google Scholar - LinkedIn - GitHub - Research Map

Research Interests: Machine Learning, Optimal Transport, Geometric Methods, Kernel Methods

My research aims to make AI more advanced, with a focus on improving its scalability, efficiency, and robustness for learning with large-scale, complex data and models from mathematical perspectives, in particular optimal transport (OT) theory. More specially, I have been pursuing two main objectives as follows:

1. Deepening OT theory:

   1. Breaking the curse-of-dimensionality for OT with projected 1D-structure, e.g., exploiting graph structure, tree system, tree structure.

   2. Going beyond Lp geometric structure for variational Wasserstein problems, e.g., leveraging functional geometric structure.

2. Leveraging OT theoretical insights for AI:

   1. Designing innovative, and efficient methodologies, e.g., for data valuation, LLM pretraining, diffusion model, generative model.

   2. Centering on applications with large-scale, complex structured data, e.g., non-Euclidean, dynamic, nonlinearity, corruption, perturbation.

News

• 20-Nov-2025, I will serve as Area Chair for ICML’26.

• 22-Oct-2025, Our paper, entitled “Tree Structure for the Categorical Wasserstein Weisfeiler-Lehman Graph Kernel” has been accepted to TMLR journal! This is joint work with Keishi Sando, and Hideitsu Hino.

• 19-Sep-2025, Our 2 papers have been accepted to NeurIPS’25.

  • An Efficient Orlicz-Sobolev Approach for Transporting Unbalanced Measures on a Graph [spotlight], joint work with Truyen Nguyen, Hideitsu Hino, and Kenji Fukumizu.

  • Tree-Sliced Entropy Partial Transport, joint work with Hoang Tran, Thanh Tran, Thanh Chu, and Tan Nguyen.

• 16-Aug-2025, I will serve as Area Chair for ICLR’26 and AISTATS’26.

• 13-July-2025, I will serve as Senior Program Committee for AAAI’26.

• 01-May-2025, Our 3 papers have been accepted to ICML’25.

  • Tree-Sliced Wasserstein Distance: A Geometric Perspective, joint work with Hoang Tran, Trang Pham, Tho Tran, Khoi Nguyen, Thanh Chu, and Tan Nguyen.

  • Tree-Sliced Wasserstein Distance with Nonlinear Projection, joint work with Thanh Tran, Hoang Tran, Thanh Chu, Trang Pham, Laurent El Ghaoui, and Tan Nguyen.

  • Scalable Sobolev IPM for Probability Measures on a Graph, joint work with Truyen Nguyen, Hideitsu Hino, and Kenji Fukumizu.

• 25-Mar-2025, Our paper, entitled “Few-Shot Object Detection via Synthetic Features with Optimal Transport” has been accepted to CVIU journal! This is joint work with Khoa Nguyen, Toan Do, Tiep Nguyen, Triet Tran, and Tam V. Nguyen.

• 23-Jan-2025, Our 3 papers have been accepted to ICLR’25.

  • SAVA: Scalable Learning-Agnostic Data Valuation, joint work with Samuel Kessler, and Vu Nguyen.

  • Distance-Based Tree-Sliced Wasserstein Distance, joint work with Hoang Tran, Khoi Nguyen, Trang Pham, Thanh Chu, and Tan Nguyen.

  • Spherical Tree-Sliced Wasserstein Distance, joint work with Hoang Tran, Thanh Chu, Khoi Nguyen, Trang Pham, and Tan Nguyen.

Vita

I received my Ph.D. in 01/2016 from Kyoto University, supervised by Marco Cuturi and advised by Akihiro Yamamoto. I worked as a postdoc at NITECH & NIMS, between 02/2016 and 08/2017. Then, I worked at RIKEN AIP between 09/2017 and 07/2021 as a researcher; and between 08/2021 and 08/2022 as a research scientist. I have joined Institute of Statistical Mathematics (ISM), Tokyo, in 09/2022, as an assistant professor.

Selected Service

• Area chair: ICML, ICLR, AISTATS

• Editor board reviewer: JMLR

• Program committee: NeurIPS

• Senior program committee: AAAI

Contact