Kohei Hatano
Kohei Hatano researches computational learning theory and online decision making. As of October 2026, he directs RIKEN’s Computational Learning Theory Team. His publications include multiple-instance learning, online load balancing and learning-based channel selection. In 2023, he coauthored a boosting-based method for converting linear threshold functions into binary decision diagrams with Yiping Tang and Eiji Takimoto. The paper connects these representations with neural-network verification.
Timeline
-
2020
Learning from informative patterns
Hatano coauthored Theory and Algorithms for Shapelet-Based Multiple-Instance Learning. The paper appeared in Neural Computation. He coauthored work on a learning setting that represents information through sets of instances and selected patterns.
-
2021
Online decisions and resources
He coauthored work on algorithms for online load balancing and energy-efficient channel selection using an upper-confidence-bound method. These projects connect learning theory with decisions that must be made over time as conditions and information change.
-
2023
Representing networks for verification
With Yiping Tang and Eiji Takimoto, Hatano developed a boosting-based method for constructing binary decision diagrams for linear threshold functions. The paper also describes combining these representations to represent a neural network. This can support the use of verification techniques.
-
As of October 2026
Learning theory at RIKEN
Hatano directs RIKEN’s Computational Learning Theory Team. The team studies online decisions, including interaction with potentially adversarial environments. Its aim is to establish strategy limits and develop robust, efficient methods through theoretical analysis.
Sources and credits
Photo: www.riken.jp; 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.