Profile

============================================================================

*** I moved my homepage to: https://tamle-ml.github.io/ ***

============================================================================

Research interests: Riemannian manifold, optimal transport/Wasserstein geometry, geometry in machine learning, topological data analysis, kernel methods, parametric optimization, metric learning.

*** News ***

• 25-Apr-2019, Python code for our ICML'19 (Safe Grid Search with Optimal Complexity) is available [Github], coded by Eugene Ndiaye.
• 22-Apr-2019, Our paper, entitled "Safe Grid Search with Optimal Complexity" has been accepted to ICML'19. This is joint work with Eugene Ndiaye, Olivier Fercoq, Joseph Salmon and Ichiro Takeuchi.
• 19-Oct-2018, Matlab code v0.1 for Persistence Fisher distance (Fisher information metric between two persistence diagrams with or without Fast Gauss Transform) is available [Github].
• 05-Sep-2018, Our paper, entitled "Persistence Fisher Kernel: A Riemannian Manifold Kernel for Persistence Diagrams" has been accepted to NeurIPS'18! (There are total 12 accepted papers from RIKEN AIP at NeurIPS'2018)
• 01-Sep-2017, Join in High-Dimensional Statistical Unit, RIKEN AIP as a postdoctoral researcher.
• 05-Apr-2017, Our proposal has been accepted by the Grants-in-Aid for Young Scientists (B) as PI (04/2017 - 03/2020).

Research Support

• 04/2017 - 03/2020, JSPS KAKENHI Grant number 17K12745 (Grants-in-Aid for Young Scientists (B) as PI).

Education

• 10/2012 - 09/2015, PhD program, Graduate school of Informatics, Kyoto University, Japan [Ranking]
  • Supervisors: Professor Marco Cuturi, Professor Akihiro Yamamoto
  • I completed my PhD program in September, 2015 and officially received my PhD degree in January, 2016.
• 09/2008 - 05/2011, MSc program, Vietnam National University, HCMC (University of Science), Vietnam
  • Rank: 1/100+
• 09/2004 - 09/2008, BSc program, Vietnam National University, HCMC (University of Science), Vietnam (honor program)
  • Rank: 3/36 (honor program) & 3/550+ (faculty)

Research Experiences

• 09/2017 - present, Postdoctoral Researcher, High-Dimensional Statistical Modeling Unit, Generic Technology Research Group, RIKEN Center for Advanced Intelligence Project (RIKEN - AIP).
  • Mentor: Professor Makoto Yamada
• 02/2016 - 08/2017, Postdoctoral Researcher, National Institute for Materials Science, Japan and Nagoya Institute of Technology, Japan.
  • Mentor: Professor Ichiro Takeuchi
• 01/2016, Associate Researcher, Nagoya Institute of Technology, Japan.
  • Mentor: Professor Ichiro Takeuchi
• 10/2015 - 12/2015, Associate Researcher, Kyoto University, Japan.
  • Mentors: Professor Akihiro Yamamoto and Professor Marco Cuturi
• 01/2014 - 02/2014, Internship student, Lear, INRIA, Grenoble, France
  • Supervisor: Professor Zaid Harchaoui
  • Subject: Ground Metric Learning
• 03/2013, Internship student, Lear, INRIA, Grenoble, France
  • Supervisor: Professor Zaid Harchaoui
  • Subject: Some methods in optimization and machine learning
• 09/2008 - 09/2012, Teaching Assistant, Researcher, and Lecturer, Vietnam National University, HCMC (University of Science), Vietnam
• 08/2010 - 02/2011, Internship student, National Institute of Informatics, Tokyo, Japan
  • Supervisor: Professor Akihiro Sugimoto
  • Project: Image Categorization using Kernel Methods
• 04/2009 - 06/2009, Internship student, Toyota Technological Institute of Nagoya, Japan
  • Supervisor: Professor Seiichi Mita
  • Project: Autonomous driving system (Module: Real time traffic sign detection)

Selected Publications

• Eugene Ndiaye, Tam Le, Olivier Fercoq, Joseph Salmon, Ichiro Takeuchi, Safe Grid Search with Optimal Complexity, to appear in International Conference on Machine Learning (ICML), US, 2019. [ArXiv/Code] (Acceptance rate: 773/3424=22.6%)
• Tam Le, Makoto Yamada, Persistence Fisher Kernel: A Riemannian Manifold Kernel for Persistence Diagrams, The 32nd Conference on Neural Information Processing Systems (NeurIPS), Canada, 2018. [PDF/Supplemental/PROJECT/POSTER] (Acceptance rate: 1011/4856=20.8%)
• Tam Le, Marco Cuturi, Unsupervised Riemannian Metric Learning for Histograms Using Aitchison Transformations, International Conference on Machine Learning (ICML), France, 2015. [VideoLecture/PDF/Supplemental/SLIDE/POSTER] (Acceptance rate: 270/1037 = 26.0%)
• Tam Le, Marco Cuturi, Adaptive Euclidean Maps for Histograms: Generalized Aitchison Embeddings, Machine Learning Journal (MLJ), 2014. [PDF/PROJECT]