论文部分内容阅读
组合Kohonen竞争学习和反向传播学习的优点,本文首次提出了复合对向-反向传播人工神经网络模型,该模型较好地体现了生物神经网络系统信息处理时的自适应、自组织、分布式存贮及并行处理等特点。它保留了反向传播网络的优点,同时较后者更易收敛,计算时间缩短,网络参数设置也更为自由。通过在临床精液检查结果分析中的成功应用,证明了该系统的有效性和可靠性。
Combined with the advantages of Kohonen competitive learning and backpropagation learning, this paper presents for the first time a composite counter-propagating artificial neural network model which can well reflect the adaptive, self-organizing and distributed information processing of biological neural network system Storage and parallel processing and other characteristics. It retains the advantages of backpropagation networks, while the latter is easier to converge, the calculation time is shortened, the network parameter settings are more free. The successful application of the results in clinical semen examination proved the effectiveness and reliability of the system.