Adaptive Critic Designs and Implementations

这个基于成就的更新项目延续了PI对一种名为“基于模型的自适应批评”的类脑智能控制新方法的研究。“它将试图统一我们对这种方法的理解，因为它与其他竞争方法有关，如随时间的暴力优化、经典方法、随时间的反向传播等。”它将基于这种方法开发和研究新的设计。此外，它将试图对现实世界工业应用中重要的稳定性和融合问题产生更深入的理解。

This Accomplishment-Based Renewal project continues the PI on a new approach to brain-like intelligent control called `Model Based Adaptive Critics.` It will attempt to unify our understanding of this approach as it relates to competing approaches such as brute-force optimization over time, classical methods, backpropagation through time, and so on. It will develop and investigate new designs based on this approach. In addition, it will attempt to generate a deeper understanding of stability and convergence issues important to real-world industry applications.