【摘要】 本文提出了将遗传算法(GA)和 δ 规则相结合,来调整一种无隐层的前向网络神经网络-函数链神经网络的参数及结构的方法,并将提出的算法用于 2bit 异或和曲面拟和问题,实验结果良好。更多还原

【Abstract】 A new method combining genetic algorithm with delta rule is proposed to adjust the parameters and structure of feed-forward function-link neural network without hidden layers. Results of the experiments on ‘XOR’ problem and curved-surface-fitting show that the proposed method has superior performance.更多还原