Reservoir computing is a paradigm that simplifies the training of recurrent neural networks. Beyond reservoir computing algorithms simulated on conventional computers, many physical implementations have been realized, where the reservoir is either composed of many physical distinct nodes, or it is folded in the dynamics of only one single node. Here I will present an experimental realization of an optoelectronic reservoir computing system, folded in the dynamics of an integrated silicon microring resonator coupled to an external optical feedback. The performance of the network is evaluated on delayed digital tasks.
Online-only seminar:
https://uibuniversitat.zoom.us/j/87008849910
