Masashi Sugiyama
Masashi Sugiyama researches statistical machine learning under limited and imperfect information. He earned his doctorate in 2001, became a University of Tokyo professor in 2014 and began directing RIKEN AIP in July 2016. He coauthored the 2017 positive-unlabeled learning paper with a non-negative risk estimator. His later work includes a 2022 book on weak supervision and research on complementary labels and test-time adaptation. He directs RIKEN’s Imperfect Information Learning Team.
- Now at
- RIKEN and University of Tokyo
- Works
- 1,005
- Citations
- 21,047
- h-index
- 67
Publication numbers from OpenAlex, as of 14 October 2026.
Timeline
-
2001–2014
A statistical learning career
Sugiyama received his doctoral degree at Tokyo Institute of Technology in March 2001. He became a University of Tokyo professor in October 2014. His research centers on statistical machine-learning theory and algorithms.
-
July 2016
Leading RIKEN AIP
He began directing RIKEN’s Center for Advanced Intelligence Project in July 2016. The Imperfect Information Learning Team studies accurate learning when large or complete datasets are unavailable.
-
2017
Learning from positive and unlabeled data
Sugiyama coauthored a method for positive-unlabeled learning with Ryuichi Kiryo, Gang Niu and Marthinus du Plessis. The paper introduces a non-negative risk estimator. It addresses a setting in which training data includes known positive cases and unlabeled examples.
-
2022–2024
Weak supervision and changing data
His team coauthored the 2022 book Machine Learning from Weak Supervision. Later papers address complementary labels and adaptation in changing test environments. These studies continue the documented focus on learning from imperfect information.
Sources and credits
Photo: www.riken.jp; Masashi Sugiyama profile photograph; creator not stated. License: Not stated by source. Source: riken.jp. Cropped to a square and resized. Ask us to remove or replace this photo.