Hi! I’m Takeshi Teshima, a third-year Ph.D. student in Machine Learning at the Unversity of Tokyo (Graduate School of Frontier Sciences), advised by Professor Masashi Sugiyama. I am a member of the Sugiyama-Yokoya-Ishida Laboratory, a Junior Research Associate at RIKEN AIP (national research center), and I am supported by Masason Foundation.
Previously I received B.Econ. from the University of Tokyo in 2017 and M.Sc. from the Unversity of Tokyo in 2019. In previous years, I have been honored to closely work with Professor Issei Sato as well as many other exceptional collaborators.
I am taking my first step toward AI for Good.
Research Interests 🔗
I’ve always been interested in generic methodology. I am currently working on developing statistical machine learning as one of the major frontiers of statistics/information technology. For more details, please visit the research introduction page.
- Causality for learning from small data.
- Transfer learning based on similarity of causal mechanisms (causal mechanism transfer).
- Incorporating causal graphical knowledge into predictive modeling (causal-graph data augmentation).
- Machine learning for social sciences / natural sciences.
- Recovering data from ceiling effects, a.k.a. censoring: clipped matrix completion.
- Machine learning methodology in general.
- Examples: classification learning from limited information (positive-unlabeled classification), robust approximate Bayesian computation, density ratio estimation, and representation power of invertible neural networks.
My CV can be found here.
Please find my current email address in the sidebar.
News and Upcoming Events 🔗
|13 Jan 2022|
Organizing a special session at StatsML Sypmposium'21 (Feb. 9-11 JST). I'm also running a count-down article event. Please join us!
|10-12 Nov 2021|
Presenting our work at IBIS 2021! On causal-graph data augmentation (No.66) on 12th Nov, and our collaborative work on 10th Nov (No.4) and 11th Nov (No.78).
|22 Oct 2021|
Released a Python package for Echelon Analysis, echelon-py!
|1 Jul 2021|
Our work on invertible neural networks has been featured in the PR magazine RIKEN!
|13 May 2021|
Our paper "Incorporating Causal Graphical Prior Knowledge into Predictive Modeling via Simple Data Augmentation" has been accepted at UAI 2021!
Older news are archived here.