An advanced neuron model for optimizing the SIREN network architecture

The paper presents a computational system based on a synchronous feedback neural network for the on-line event processing of a photon counting intensified CCD. The project is based on a neuron model that improves the one already defined for the SIgnal REcognition Network (SIREN) system and on a suitable network architecture. As far as the neuron is concerned, a new equation for network dynamics is envisaged, that allows to reduce the number of cycles needed for event identification. As far as the architecture is concerned, we focus on the definition of a network having less neurons than SIREN while maintaining the same performance. An hypothesis of network architecture and a performance comparison between the two neural models are given in the paper.