Nonlinear neural network oscillator

A nonlinear oscillator, comprising:

• a neural network having at least one output for outputting a one dimensional vector, the neural network including a plurality of layers, including an input layer, an output layer, and at least one hidden layer, each of the layers including at least one processing element, wherein an output of each said processing element of the input layer and the hidden layer is connected to an input of at least one processing element of another layer, wherein the input layer has an input coupled to the at least one output and includes an analog delay line means having a plurality of taps each of which outputs a time-delayed sample of the one dimensional output vector, each of the taps being connected to each one of the processing elements of the at least one hidden layer for providing a sample at one of the taps of at least one of said m delay lines.

A vector converter comprising at least one output providing a first one dimensional vector in response to a first input vector and providing a second one dimensional vector in response to a second input vector, wherein the first one dimensional output vector differs from the second one dimensional vector by an amount that is greater than an amount that the first input vector differs from the second input vector, the vector converter comprising:

• a neural network including an input layer for receiving the input vectors, an output layer for outputting the at least one output, and at least one hidden layer, each said hidden layer including at least one processing element, wherein an output of each said processing element of the input layer and the hidden layer are connected to an input of at least one processing element of another layer, and wherein a strength of a connection between two processing elements is a function of an associated weight, each said weight being trained on a deterministic sequence of data from a broadband chaotic source.