人工智能研究的物理学途径

THE PHYSICAL APPROACH TO ARTIFICIAL INTELLIGENCE

近年来 ,人工智能研究的物理学途径已越来越受到人们的关注 .这一途径的实现方式主要有两种 :一是把已有的物理理论和方法直接用于研究计算系统 ,以达到对计算系统的性质和规律的认识 ;另一是依据某些计算系统与物理系统的相似性 ,把计算问题转换为对应的物理问题 ,再用物理方法建构有效的算法加以求解 .文章着重介绍运用物理学途径研究计算系统的相变和解决多自主体系统的目标满足问题所取得的进展 ,并指出了这一途径的优点和应该注意的问题 .

The physical approach to artificial intelligence has attracted increasing attention in recent years.This approach is implemented in two ways:the first is by using physics directly as a framework for understanding the behavior of computational systems,the second is by using physics oriented methods to construct effective algorithms for resolving computational problems according to the similarities between computational and physical problems.In this paper we describe and analyze the advantages of the application of the physical approach to phase transitions in computational systems and cooperative goal satisfaction in multi agent systems.