Learning programs for decision and control

Introduces learning programs, an approximate dynamic programming (ADP) or otherwise named neural dynamic programming (NDP) algorithm developed and tested by the authors. We first introduce the basic framework of our learning programs, the associated learning algorithms, and then extensive case studies to demonstrate the effectiveness of our learning programs. This is probably the first time that neural dynamic programming type of learning algorithms has been applied to complex, real life continuous state problems. Until now, reinforcement learning (another learning approach for approximate dynamic programming) has been mostly successful in discrete state space problems. On the other hand, prior NDP based approaches to controlling continuous state space systems have all been limited to smaller, or linearized, or decoupled problems. Therefore the work presented here compliments and advances the existing literature in the general area of learning approaches in approximate dynamic programming.