Kazuyuki Samejima
Tamagawa University, Tokyo University of Agriculture and Technology
Papers
4
Total Citations
186
H-Index
4
About
Kazuyuki Samejima is a leading figure in computational neuroscience and robotics, best known for bridging reinforcement learning theory with biological and artificial systems. His seminal 2007 work, "Efficient reinforcement learning: computational theories, neuroscience and robotics" (91 citations), synthesizes frameworks that unify machine learning algorithms with neural mechanisms of decision-making, offering a roadmap for designing adaptive agents. Earlier, Samejima pioneered methods for real-world reinforcement learning, notably in his 1999 paper "Adaptive internal state space construction method for reinforcement learning of a real-world agent" (57 citations), which introduced algorithms for autonomously building state representations from high-dimensional sensory data—a critical step for deploying RL in physical robots. His research on motion sequence learning, including the 2006 study "Symbolization and imitation learning of motion sequence using competitive modules" (30 citations), extends the MOSAIC architecture to enable robots to recognize and imitate complex movements through modular prediction models. Samejima’s work has profoundly impacted both computational theory and embodied AI, demonstrating how biological principles of learning and control can inspire more efficient, adaptive robotic systems. His contributions remain foundational for researchers exploring the intersection of neuroscience, machine learning, and autonomous robotics.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4Adaptive State Space Formation Method for Reinforcement Learning.8 citations · 1999