Photonic delay dynamics emulating spatio-temporal networks: From chimera states to photonic artificial intelligence

Delay dynamics are well known for their infinite dimensional phase space. An analogy with space-time dynamics was proposed 20 years ago from a theoretical perspective. In this talk, we will develop such an analogy from two particular point of views recently explored through experimental investigations in photonic, with optoelectronic nonlinear delay oscillators. Emerging and self sustained spatio-temporal patterns, known as chimera states in network of oscillators, will be first reported and analyzed in such purely temporal, but high-dimensional enough, dynamics. The conditions for their occurrence will be described and analyzed with respect to the characteristic operating parameters of the delay dynamics, and a highly multi-stable organization of the parameter space will be reported. Second, a novel application of delay dynamics in photonic will be reported, consisting in a brain-inspired information processing system. Again, space-time properties will be emphasized, showing how a delay dynamics can efficiently emulate a complex ensemble which acts as a network of neurons capable to process information. An experimental speech recognition test will be reported with unprecedented processing speed, up to 1 million words/second classification rate, moreover with very low word error rate.