Jun Namikawa
Papers
7
Total Citations
237
H-Index
6
About
Jun Namikawa is a pioneering researcher at the intersection of computational neuroscience and robotics, whose work fundamentally explores how dynamic neural networks can model and generate complex, adaptive behaviors. His primary research areas include neural network models for time series prediction, robot learning through human tutoring, and the neurodynamic basis of spontaneous action. Namikawa’s major contribution lies in developing novel recurrent neural network architectures that can infer time-dependent stochastic properties from fluctuating data, enabling robots to learn not just the mean but also the variance of sensory inputs—a crucial step for robust, human-like interaction. His most-cited work, “Learning to Reproduce Fluctuating Time Series” (75 citations), introduces a dynamic model that extracts hidden stochastic structures, directly applied to robot learning via tutoring. In “Codevelopmental Learning Between Human and Humanoid Robot” (50 citations), he demonstrates how hierarchical neural networks, inspired by the human parietal cortex, facilitate interactive learning. His theoretical article “A Neurodynamic Account of Spontaneous Behaviour” (40 citations) proposes that deterministic chaos in cortical dynamics generates action sequences, linking neural mechanisms to behavioral variability. Through these achievements, Namikawa has advanced our understanding of how robots can learn goal-directed actions and cooperative behaviors, bridging cognitive science and artificial intelligence.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3A Neurodynamic Account of Spontaneous Behaviour40 citations · 2011
- 4
- 5
- 6
- 7Reinforcement Learning Algorithm with CTRNN in Continuous Action Space3 citations · 2006