【摘要】 为克服神经网络软件仿真实时性差、并行处理能力弱等缺点,提出了采用电路的方法实现秀丽线虫的接触感知神经网络,模拟秀丽线虫的回撤行为。其中所有参数由实数编码遗传算法训练秀丽线虫接触感知神经网络模型所得参数转化而来。通过Hspice仿真器进行仿真,Hspice仿真结果和秀丽线虫接触感知神经网络模型的数值仿真结果相符,验证了该电路的有效性和正确性。更多还原

【Abstract】 To overcome bad real time effect and weak parallel processing ability of neural network simulated by software,using analog circuit to implement a neural network for touch sensitivity in Caenorhabditis elegans was proposed,and the nematode’s withdrawal behaviour was reproduced with analog circuit at the same time.All parameters included in the neural network implemented by analog circuit were converted from the parameters acquired by using the real-coded genetic algorithm to train the neural network for touch sensitivity in Caenorhabditis elegans.The analog circuit was simulated by Hspice.The results of circuit simulated by Hspice were consistent with the numerical results obtained from the neural network model,which showed the validity and the correctness of analog circuit.更多还原