Hidenori Itaya

About

Hidenori Itaya is a researcher focused on advancing the interpretability of deep reinforcement learning (DRL) systems. His primary research areas include explainable artificial intelligence (XAI), attention mechanisms, and agent decision-making analysis. Itaya’s major contributions center on developing visual explanation methods that demystify the "black-box" nature of DRL agents, particularly in complex environments like games and robot control. His most-cited work, "Visual Explanation using Attention Mechanism in Actor-Critic-based Deep Reinforcement Learning" (2021), has garnered 22 citations for proposing a novel approach that uses attention mechanisms to visualize why an agent selects specific actions, thereby enhancing transparency. Building on this, his 2024 paper "Mask-Attention A3C: Visual Explanation of Action–State Value in Deep Reinforcement Learning" (3 citations) further refines these techniques by masking irrelevant features to highlight key state-action values. These achievements are notable for bridging the gap between high-performing DRL models and human understanding, a critical step for deploying AI in safety-sensitive domains. Itaya’s work empowers researchers and practitioners to trust and debug autonomous systems, making his contributions both technically innovative and practically impactful.

Research Focus

Key Achievements

Top Papers

1. 1
   Visual Explanation using Attention Mechanism in Actor-Critic-based Deep Reinforcement Learning

   22 citations · 2021
2. 2
   Mask-Attention A3C: Visual Explanation of Action–State Value in Deep Reinforcement Learning

   3 citations · 2024

Key Collaborators

Contact & Links