Learning and Forgetting in Generalized Brain-state-in-a-box (BSB) Neural Associative Memories

A novel neural network for associative memory via dynamical systems

“…The conventional approach is to synthesize a connection matrix with the given prototype pattern vectors so that such vectors are stored as asymptotically stable equilibrium points. This approach has been studied extensively and some of its advantages and disadvantages are discussed by Żak et al [19]. In particular, problems related to memory capacity, stability (the basin of attraction of each memory pattern) and spurious memory have attracted a lot of attentions.…”