Adaptive evolutional learning method of neural networks using genetic algorithms under dynamic environments

摘要

Backpropagation learning and genetic algorithms are widely known for their superior adaptation capability by imitating mechanisms of a living thing. However, most studies in this field have been developed under static environments. Once input-output patterns change, the trained network under static environments should start training from the initial state. On the contrary, if their algorithms have a sufficient adaptive ability under dynamic environments, they can work like a living thing's evolutionary process. We propose an adaptive evolutional learning method of neural networks using genetic algorithms, which can perform effective learning under dynamic environments.