本文在研究了ART算法、Field Theory和ARTMAP等算法的基础上提出了FTART算法,它将自适应谐振理论、域理论等的优点有机结合.以样本在实例空间中出现的概率为启发信息修改网络中的分类,并采用了不同于其它算法的解决样本间的冲突和动态扩大分类区域的方法.测试结果表明,FTART算法具有比较好的性能. Based on the research of algorithms such as ART, Field Theory and ARTMAP, this paper proposes FTART algorithm, which combines the advantages of adaptive resonance theory and domain theory, etc. The probability that the sample appears in the instance space is the inspiration of information modification network And uses a different method from other algorithms to solve the conflict between samples and dynamically expand the classification area.The test results show that the FTART algorithm has better performance.