Related Experiment Videos
Learning model for coupled neural oscillators.
1Department of Physics, Biology and Informatics, Faculty of Science, Yamaguchi University, Japan.
Network (Bristol, England)
|September 25, 1999
Summary
This study introduces learning models for coupled neural oscillators, enabling them to adapt intrinsic frequencies and coupling strengths. This research aids in understanding motor command learning in biological systems.
Area of Science:
- Computational Neuroscience
- Robotics
- Machine Learning
Background:
- Motor commands in living systems are often generated by coupled neural oscillators.
- Proper coordination requires precise intrinsic frequencies and coupling strengths.
Purpose of the Study:
- To propose novel learning models for coupled neural oscillators.
- To enable acquisition of desired intrinsic frequencies and coupling weights.
- To understand the learning mechanisms of motor commands.
Main Methods:
- Development of learning rules for coupled neural oscillators.
- Utilizing desired phase patterns or evaluation functions for learning.
- Computer simulations, including adaptive control of a hopping robot.
Main Results:
- The proposed learning rules effectively acquire desired frequencies and coupling weights.
- Demonstrated adaptive control capabilities in a simulated hopping robot.
- The learning rule is simple, resembling a Hebbian rule.
Conclusions:
- The developed learning models offer a viable method for controlling coupled neural oscillators.
- This work provides insights into the biological learning of motor commands.
- Further studies on these models can advance understanding of neural control systems.