Discrete network implementation of Freeman's model of the olfactory system

This paper presents results concerning a full digital implementation of the KII and KIII model proposed by Freeman to explain the chaotic behavior of the mammalian olfactory cortex (1995). The environment used for the task is a commercial neural network package. The immediate goal is to offer a computational efficient alternative to Runge-Kutta methods of solving the system's differential equations. To accomplish the goal the differential equations' impulse response were sampled and decomposed over the gamma basis. Both models, KII and KIII, are presently working as associative memories, where stored patterns can be recalled from an input pattern.