Takuya Isomura

RIKEN Center for Brain Science, The University of Tokyo

Papers

Total Citations

H-Index

About

Takuya Isomura is a pioneering researcher at the intersection of computational neuroscience, robotics, and Bayesian inference. His work centers on the free-energy principle and active inference, frameworks that explain how biological and artificial systems perceive and act under uncertainty. Isomura’s major contribution is bridging theoretical neuroscience with practical robotics: his 2021 paper, “Kalman filters as the steady-state solution of gradient descent on variational free energy” (3 citations), reveals that the classic Kalman filter emerges naturally from minimizing variational free energy—a foundational insight linking control theory to brain-inspired learning. Building on this, his 2024 work, “Real-World Robot Control Based on Contrastive Deep Active Inference With Demonstrations” (2 citations), demonstrates how robots can learn complex tasks by combining deep learning with active inference, using human demonstrations to accelerate learning. This approach addresses a critical gap between human-like perception and robotic action. Isomura’s research has been cited in top venues including *Neural Computation* and *IEEE Robotics and Automation Letters*, and his 1985 early work on robot control programming (2 citations) shows a lifelong commitment to autonomous systems. His contributions are shaping next-generation robots that learn and adapt like living organisms.