Cover story: Han YN, Xiang SY, Song ZW, Gao S, Guo XX et al. Pattern recognition in multi-synaptic photonic spiking neural networks based on a DFB-SA chip. Opto-Electron Sci 2, 230021 (2023).

Neuromorphic computing that inspired by biological intelligence is expected to be more powerful and less power-consuming. The spike event observed in neural system demonstrates spatio-temporal encoding and event-driven properties for information processing. In recent years, optical platform has been widely concerned as a promising candidate for computing acceleration due to the ultrahigh speed, low latency and wideband. Hence, photonic spiking neural network (SNN) serves as an ultra-fast and energy-efficient platform for high-performance neuromorphic computing, and attracts increasing attentions. As the basic element in neural network, the spiking neuron can be emulated via a two-section semiconductor laser (namely, with separated gain and saturable absorption region), which is capable of mimicking neuron-like responses and perform nonlinear calculations like neurons. The integrated architecture provides a reliable optical neuromorphic computing architecture scheme due to the potential with compact size and high reliability. However, the strong nonlinearity of SNN and the training difficulties caused by hardware constraints put forward an urgent need for efficient SNN algorithms, and the collaborative computing of algorithms and hardware platforms has become a new challenge.