论文部分内容阅读
ＦＴＡＲＴ（ｆｉｅｌｄｔｈｅｏｒｙ－ｂａｓｅｄＡＲＴ）算法结合了ＡＲＴ（ａｄａｐｔｉｖｅｒｅｓｏｎａｎｃｅｔｈｅｏｒｙ）算法、ＡＲＴＭＡＰ算法、域理论的思想，以样本在实例空间中出现的概率为启发信息修改学习中生成的分类，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．本文在对ＦＴＡＲＴ算法的研究的基础上进行了改进，使算法在学习连续函数的映射时更加有效．同时给出了算法的测试结果和对测试结果的分析，测试表明，ＦＴＡＲＴ算法在模式识别和连续函数映射的学习方面具有比较好的性能
The FTART (fieldtheory-basedART) algorithm combines ART (Adaptive Resonancetheory) algorithm, ARTMAP algorithm and domain theory, and uses the probabilities that the samples appear in the instance space to modify the classification generated in the learning as the inspiration information, and adopts a solution different from other algorithms Conflict between samples and dynamic ways to expand classified areas. This paper improves on the research of FTART algorithm and makes the algorithm more effective when learning the mapping of continuous functions. At the same time, the algorithm test results and the analysis of the test results are given. The tests show that the FTART algorithm has better performance in pattern recognition and continuous function mapping