Masashi Sugiyama

A new approach to reinforcement learning addresses delayed rewards using bagged feedback.

Examining how noise in pre-training data impacts model performance.

Learn how computers make decisions through Reinforcement Learning techniques and processes.

FedLESAM tackles data challenges in federated learning for better model performance.

Investigating how small errors in training data enhance AI-generated content.

New methods improve decision-making in multiplayer scenarios using preference-based feedback.

Exploring efficient methods for accurate log-concave sampling in complex data environments.

Introducing CLIPFit, a method for efficient fine-tuning of Vision-Language Models.

Discover how parallel sampling methods transform data analysis in scientific research.