Adaptive genetic programs via reinforcement learning
Keith L. Downing · 2001
Reinforced Genetic Programming (RGP) enhances standard tree-based genetic programming (GP) [7] with reinforcement learning (RL) [11]. Essentially, leaf nodes of GP trees become monitored action-selection points, while the internal nodes form a decision tree for classifying the current state of the problem solver. Reinforcements returned by the problem solver govern both fitness evaluation and intra-generation learning of the proper actions to take at the selection points. In theory, the hybrid RGP system hints of mutual benefits to RL and GP in controller-design applications, by, respectively, providing proper abstraction spaces for RL search, and accelerating evolutionary progress via Baldwinian or Lamarckian mechanisms. In practice, we demonstrate RGP's improvements over standard GP search on maze-search tasks