论计算智能及其应用

COMPUTATIONAL INTELLIGENCE AND ITS APPLICATION

文中论述了计算智能（ｃｏｍｐｕｔａｔｉｏｎａｌｉｎｔｅｌｉｇｅｎｃｅ，ＣＩ）这一新兴研究开发领域的形成背景、内容、方法特征及其与人工智能（ＡＩ）的区别与联系．指出主要的ＣＩ算法，如神经网络的训练算法、遗传算法、演化算法等，实质上是一类准元算法，适合于解决那些难以建立确定性数学／逻辑模型，或不存在可形式化模型的问题．文中还提出并讨论了演化计算的生态学模型及一般算法．进一步论述了ＣＩ＋ＡＩ多元复合协同计算系统结构及其在非线性随机动态或混沌系统分析，例如市场价格变化趋势分析，或ＮＰ难问题求解等方面的应用前景．

This paper gives a survey of the background, intention, contents, and methodological peculiarities of Computational Intelligence (CI), one of the flourishing advanced R & D area. The essential differences and associations between CI and the traditional AI are discussed as well. It is argued that the main CI algorithms, such as the training algorithms for neural networks, the genetic algorithms, the evolutionary algorithms, and so on, should be considered as a category of quasi meta algorithm. They are especially good for solving problems that are very hard, if not impossible at all, to have deterministic mathematical or logic models. An ecological model of evolutionary computation and a general algorithm are also proposed and discussed. The application principles and perspectives of CI+AI hybrid synergetic computing systems in analyzing non linear, stochastic, dynamic or chaos systems, such as the trend analysis of the market pricing, solving NP hard problems, etc., are further discussed with sample examples.