Learning as a Nonlinear Line of Attraction for Pattern Association, Classification and Recognition [unknown]

2006
The human brain memorizes information using a dynamical system made o f interconnected neurons. Retrieval o f information is achieved in an associative sense, starting from an arbitrary state that might be an encoded visual representation; the brain activity converges to a stable state in which the brain remembers. Associative memory can be modeled using a recurrent network, in which the stored memories are represented by the dynamics o f the network convergence. Development o f a mathematical
more » ... o f a mathematical model for learning a nonlinear line o f attraction is presented in this dissertation, in contrast to the conventional recurrent neural network model in which the memory is stored in an attractive fixed point at discrete location in state space. A nonlinear line o f attraction is the encapsulation o f attractive fixed points scattered in state space as an attractive nonlinear line, describing patterns with similar characteristics as a family o f patterns. It is usually o f prime imperative to guarantee the convergence o f the dynamics o f the recurrent network for associative learning and recall. We propose to alter this picture. That is, if the brain remembers by converging to the state representing familiar patterns, it should also diverge from such states when presented by an unknown encoded representation o f a visual image. The conception o f the dynamics o f the nonlinear line attractor network to operate between stable and unstable states is the second contribution in this dissertation research. These criteria can be used to circumvent the plasticity-Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
doi:10.25777/rtv9-tw07 fatcat:j4gq7gmt7vcrlkrbrft7joej4a