FTART2神经网络及其规则抽取研究

本文提出了一种基于域理论的自适应谐振神经网络算法ＦＴＡＲＴ２，算法将自适应谐振理论和域理论的优点有要结合，不需人为设置隐层神经元，学习速度快，精度高。此外，本文不提出了一种从ＦＴＡＲＴ２网络中抽取符号规则的方法。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，可以很好地描述了ＦＴＡＲＴ２网络的性能。

RESEARCH OF FTART2 NEURAL NETWORK AND ITS RULE EXTRACTION

In this paper, a new neural learning algorithm named FTART2, which organically combines the advantages of Adaptive Resonance Theory and Field Theory, is proposed. FTART2 overcomes the disadvantage of traditional feed forward neural networks, which need user to set up hidden units, and achieves fast learning speed and strong generality. Moreover, We propose a method to extract symbolic rules from trained FTART2 network. Experimental results show that the rules extracted through this method are comprehensible and accurate, and can commendably describe the function of original neural network.