Neuromorphic networks for signal and information processing have acquired recently a renewed interest considering the more and more complex tasks that have to be solved automatically in current applications: speech recognition, dynamic image correlation, rapid decision processing integrating a plurality of information sources, behavior optimization, etc… Several types of neuromorphic networks do exist and, among them, the spiking type (SNN), that is, the one closest in behavior to the natural cortical neurons. SNN are the ones who seem to be able to offer a best energy efficiency and thus offer scalability. Several demonstration have been made in this domain with electronic circuits and more recently with photonic circuits. For these, the dense integration potential of silicon photonics is a real advantage to create complex and highly connected circuits susceptible to lead to complete demonstrations. The PhD goal is to exploit a photonics spiking neuromorphic network architecture based on pulsed (Q-switched) lasers interconnected by a dense and reconfigurable optical network on chip mimicking the synaptic weights. A complete laser, neuron then circuit model is expected with, in the end, the practical demonstration of an application in mathematical data processing (to be defined).