Neuromorphic photonic computing has emerged as a competitive computing paradigm to overcome the bottlenecks of the von-Neumann architecture. Linear weighting and nonlinear spike activation are two fundamental functions of a photonic spiking neural network (PSNN). However, they are separately implemented with different photonic materials and devices, hindering the large-scale integration of PSNN. Here, we propose, fabricate and experimentally demonstrate a photonic neuro-synaptic chip enabling the simultaneous implementation of linear weighting and nonlinear spike activation based on a distributed feedback (DFB) laser with a saturable absorber (DFB-SA). A prototypical system is experimentally constructed to demonstrate the parallel weighted function and nonlinear spike activation. Furthermore, a four-channel DFB-SA laser array is fabricated for realizing matrix convolution of a spiking convolutional neural network, achieving a recognition accuracy of 87% for the MNIST dataset. The fabricated neuro-synaptic chip offers a fundamental building block to construct the large-scale integrated PSNN chip.

Spiking neural networks (SNNs) utilize brain-like spatiotemporal spike encoding for simulating brain functions. Photonic SNN offers an ultrahigh speed and power efficiency platform for implementing high-performance neuromorphic computing. Here, we proposed a multi-synaptic photonic SNN, combining the modified remote supervised learning with delay-weight co-training to achieve pattern classification. The impact of multi-synaptic connections and the robustness of the network were investigated through numerical simulations. In addition, the collaborative computing of algorithm and hardware was demonstrated based on a fabricated integrated distributed feedback laser with a saturable absorber (DFB-SA), where 10 different noisy digital patterns were successfully classified. A functional photonic SNN that far exceeds the scale limit of hardware integration was achieved based on time-division multiplexing, demonstrating the capability of hardware-algorithm co-computation.

Dendrites, branches of neurons that transmit signals between synapses and soma, play a vital role in spiking information processing, such as nonlinear integration of excitatory and inhibitory stimuli. However, the investigation of nonlinear integration of dendrites in photonic neurons and the fabrication of photonic neurons including dendritic nonlinear integration in photonic spiking neural networks (SNNs) remain open problems. Here, we fabricate and integrate two dendrites and one soma in a single Fabry–Perot laser with an embedded saturable absorber (FP-SA) neuron to achieve nonlinear integration of excitatory and inhibitory stimuli. Note that the two intrinsic electrodes of the gain section and saturable absorber (SA) section in the FP-SA neuron are defined as two dendrites for two ports of stimuli reception, with one electronic dendrite receiving excitatory stimulus and the other receiving inhibitory stimulus. The stimuli received by two electronic dendrites are integrated nonlinearly in a single FP-SA neuron, which generates spikes for photonic SNNs. The properties of frequency encoding and spatiotemporal encoding are investigated experimentally in a single FP-SA neuron with two electronic dendrites. For SNNs equipped with FP-SA neurons, the range of weights between presynaptic neurons and postsynaptic neurons is varied from negative to positive values by biasing the gain and SA sections of FP-SA neurons. Compared with SNN with all-positive weights realized by only biasing the gain section of photonic neurons, the recognition accuracy of Iris flower data is improved numerically in SNN consisting of FP-SA neurons. The results show great potential for multi-functional integrated photonic SNN chips.

We proposed and experimentally demonstrated a simple and novel photonic spiking neuron based on a distributed feedback (DFB) laser chip with an intracavity saturable absorber (SA). The DFB laser with an intracavity SA (DFB-SA) contains a gain region and an SA region. The gain region is designed and fabricated by the asymmetric equivalent π-phase shift based on the reconstruction-equivalent-chirp technique. Under properly injected current in the gain region and reversely biased voltage in the SA region, periodic self-pulsation was experimentally observed due to the Q-switching effect. The self-pulsation frequency increases with the increase of the bias current and is within the range of several gigahertz. When the bias current is below the self-pulsation threshold, neuronlike spiking responses appear when external optical stimulus pulses are injected. Experimental results show that the spike threshold, temporal integration, and refractory period can all be observed in the fabricated DFB-SA chip. To numerically verify the experimental findings, a time-dependent coupled-wave equation model was developed, which described the physics processes inside the gain and SA regions. The numerical results agree well with the experimental measurements. We further experimentally demonstrated that the weighted sum output can readily be encoded into the self-pulsation frequency of the DFB-SA neuron. We also benchmarked the handwritten digit classification task with a simple single-layer fully connected neural network. By using the experimentally measured dependence of the self-pulsation frequency on the bias current in the gain region as an activation function, we can achieve a recognition accuracy of 92.2%, which bridges the gap between the continuous valued artificial neural networks and spike-based neuromorphic networks. To the best of our knowledge, this is the first experimental demonstration of a photonic integrated spiking neuron based on a DFB-SA, which shows great potential to realizing large-scale multiwavelength photonic spiking neural network chips.

脑科学与类脑研究是国际必争战略性前沿。人工智能与深度学习的飞速发展对算力提出了迫切需求。而传统的冯诺依曼架构，由于存算分离导致功耗墙和内存墙，摩尔定律也逐渐放缓。光神经拟态计算充分融合高速光通信、光互连、光集成、硅基光电子与神经拟态计算的特点，具有超高速、大带宽、多维度等优势，在高性能计算、人工智能领域有广阔的应用前景，是突破后摩尔时代传统微电子计算极限极具竞争力的方案。本文回顾了国内外主要研究团队在光神经元、光突触、光神经网络的理论、算法及器件方面的工作，并提出了展望。

The explosive growth of data and information has motivated various emerging non-von Neumann computational approaches in the More-than-Moore era. Photonics neuromorphic computing has attracted lots of attention due to the fascinating advantages such as high speed, wide bandwidth, and massive parallelism. Here, we offer a review on the optical neural computing in our research groups at the device and system levels. The photonics neuron and photonics synapse plasticity are presented. In addition, we introduce several optical neural computing architectures and algorithms including photonic spiking neural network, photonic convolutional neural network, photonic matrix computation, photonic reservoir computing, and photonic reinforcement learning. Finally, we summarize the major challenges faced by photonic neuromorphic computing, and propose promising solutions and perspectives.

We propose and demonstrate experimentally and numerically a network of three globally coupled semiconductor lasers (SLs) that generate triple-channel chaotic signals with time delayed signature (TDS) concealment. The effects of the coupling strength and bias current on the concealment of the TDS are investigated. The generated chaotic signals are further applied to reinforcement learning, and a parallel scheme is proposed to solve the multiarmed bandit (MAB) problem. The influences of mutual correlation between signals from different channels, the sampling interval of signals, and the TDS concealment on the performance of decision making are analyzed. Comparisons between the proposed scheme and two existing schemes show that, with a simplified algorithm, the proposed scheme can perform as well as the previous schemes or even better. Moreover, we also consider the robustness of decision making performance against a dynamically changing environment and verify the scalability for MAB problems with different sizes. This proposed globally coupled SL network for a multi-channel chaotic source is simple in structure and easy to implement. The attempt to solve the MAB problem in parallel can provide potential values in the realm of the application of ultrafast photonics intelligence.