从存在到演化:人工智能研究范式的转变及其哲学思想  

From Existence to Evolution: the Change of Paradigm of Artificial Intelligence Research and Its Philosophic Idea

作　　者：刘仲蓓[1] 孙婧[2] 

机构地区：[1]南京大学哲学系 [2]东南大学人文学院

出　　处：《系统辩证学学报》2001年第2期71-76,共6页Journal of Systemic Dialectics

摘　　要：人工智能的研究从符号主义 (传统 AI)和联结主义 (人工神经网络 ANN)两个方向展开。 ANN的复兴在于霍普菲尔德将分布系统、热力学统计物理思想引入 ANN建模研究中 ,实现了 ANN研究从线性到非线性、从集中到分布、从动力学范式到热力学范式的转变。行为 AI和演化计算则基于达尔文进化论思想 ,从与环境实时互动联系的角度 ,实现对传统 AI局限的突破。从哲学思想上看 ,二者的共同点就在于其研究指导思想从存在走向了演化 ,贯穿了唯物辩证法普遍联系和永恒发展的思想。Artificial intelligence research developed by the way of symbolicism (tradition AI) and connectionism (ANN). Revival of ANN depended on that Hopfield carried the thought of distribution system and thermodynamics into the process of model building of ANN. It brought out the change of ANN research paradigm from linear to non linear, concentration to distribution, dynamics to thermodynamics. Behavior AI and evolution calculation based on the theory of evolution, broke out the limit of tradition AI, by the way of inter connection with environment. From the view of philosophic idea, the common point of the two ways is that the direct idea of AI research changed from existence to evolution. It is full of the materialist dialectics idea of general connection and eternal development. Development of modern AI gives an important way for researching the origin of identification.\;

关 键 词：传统AI 人工神经网络 分布 范式 存在 演化 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]