论文部分内容阅读
通过引入不同概率的双峰无规神经激活阈分布,来考虑对神经网络“记忆”恢复特性的影响,结果表明即使储存模式数超过孤立Hopfield模型的临界值α_c时系统仍然能成功地恢复储存信息。
By introducing the bimodal random neural activation threshold distribution with different probabilities, the effect on neural network “memory” recovery characteristics is considered. The results show that the system can still successfully restore stored information even if the number of stored modes exceeds the critical value α_c of the isolated Hopfield model .