Smooth recollection of a pattern sequence by nonmonotone analog neural networks
Mizuki Morita · 2002
An analog neural network model with continuous-time dynamics is presented which can memorize almost arbitrary pattern sequences. Although this model does not have any particular delay circuits or synchronizing mechanisms, the state of the network changes gradually from pattern to pattern in recalling. Numerical experiments show that the trajectory along the stored sequence can be regarded as a dynamic attractor with a large basin.>