在自适应谐振理论和域理论的基础上提出了一种基于域理论的自适应谐振神经网络算法FTART .该算法将自适应谐振理论和域理论的优点有机结合 ,采用了独特的解决样本间冲突和动态扩大分类区域的方法 ,不需人为设置隐层神经元 ,学习速度快 ,精度高 此外 ,还提出了一种从FTART网络中抽取符号规则的方法 ,即基于统计的产生 -测试法 ,实验结果表明 ,使用该方法抽取出的符号规则可理解性好、预测精度高 ,可以很好地描述FTART网络的性能 Based on adaptive resonance theory and domain theory, an adaptive resonance neural network algorithm FTART based on domain theory is proposed, which combines the advantages of adaptive resonance theory with domain theory and adopts a unique solution to inter-sample conflicts In addition, a new method of extracting symbolic rules from FTART networks is proposed, that is, the statistics-based generation-test method, the experiment The results show that the symbol rules extracted by this method have good comprehensibility and high prediction accuracy, and can well describe the performance of FTART networks