增量式IHMCAP算法的研究及其应用

RESEARCH AND APPLICATION OF THE INCREMENTAL ALGORITHM IHMCAP

增量式IHMCAP算法采用适合于混合型学习的FTART神经网络，成功解决了符号学习与神经网络学习精度之间的均衡性问题．该算法还具有较强的增量学习能力，在给系统增加新的示例时，不用重新生成已有判定树和神经网络，只需进行一遍增量学习即可调整原结构以提高学习精度，效率高，速度快．

The incremental algorithm IHMCAP successfully settles the dilemma of learning accuracy proportion between symbolic and neural parts by adopting neural network algorithm FTART which fits for hybrid learning. It also has the incremental learning abllity of adjusting old structures to improve learning accuracy by one time learning instead of rebuilding the decision tree and the neural networks when new examples are provided.