Oscillatory state machine

In computational theory, the search for finite description of data stream is moving from less computationally capable classes to more capable and optimal description or representation of the data series in question: subword string, tree representation, and finitary machine representation. In dynamical system theory, the system evolves through different hierarchical levels to cope with its environment. Evolution of the dynamical system: fixed point to limit cycle to torus and to strange attractor. In recent studies, it has been found that recurrent neural networks are capable in simulating regular grammars. Our interest is in the study of nonlinear oscillatory neural network, the oscillatory state machine (OSM), where its architecture is similar to recurrent neural networks. We are interested in the dynamics of formation and cognition of collective properties. We present an array of modified van der Pol nonlinear oscillators, which, through the device of “strange attractors”, promises to carry out very complex functional repertoires with very simple hardware and to behave as a versatile information processor