论文部分内容阅读
This paper further studies the ability of the associate learning and self-correcting in a memristive artificial neural network (ANN).Different from the existing models, the present ANN contains the multiply-threshold neurons, the discrete charge-controlled memristors, and a new learning law named the max-input-feedback (MIF).We shall demonstrate the processes of the associative learning and associative correcting via a modified Pavlov experiment where more conditioning factors are considered.We also make some comparisons of MIF with spike-timing- dependent plasticity and back-propagation and show that MIF learning law is suitable to fast learning.