利用遗传算法进一步优化CBR案例推理模型

摘要

基于案例推理是近年来人工智能领域内兴起的一种推理技术,推理指标特征的选择一直是该技术的热点和难点.为了在指标选择过程中得到较优的特征子集,本文结合灰色关联度分析和遗传算法优化特征的遴选过程,将灰色关联分析结果作为遗传算法的初始种群进行启发式搜索,一方面可以得到更优特征组合,另一方面有效减少了遗传算法的进化代数,提高了遗传算法运行效率.并基于此,提出优化的GA-CBR案例推理模型.实验结果表明,该模型有效提高了CBR预测准确性.%Case based reasoning (CBR) is a kind of reasoning technology in the field of artificial intelligence in recent years. Feature selection is the hot topic and the difficulty in CBR technology. In order to get optimal feature subset in feature selection process, this paper combines gray correlation degree analysis with genetic algorithm (CA) , taking the gray correlation analysis result as the initial population for GA heuristic search, and proposes a new GA-CBR case reasoning optimization model. The test results show this model is effective to improve the CBR forecast accuracy.