LAUSR

As one of the most promising methods in the next generation of neuromorphic systems, memristor-based spiking neural networks (SNNs) show great advantages in terms of power efficiency, integration density and biological plausibility. However, because of the non-differentiability of discrete spikes, it is difficult to train SNNs with gradient descent and error backpropagation online. In this paper, we propose an improved training algorithm for multi-layer memristive SNN (MSNN) with three methods spontaneously, supporting in-situ learning in hardware. First, temporal order encoding is applied to generate different pulse trains in neurons. Second, a simplified homeostasis is realized by the activation state and refractory period to regulate hidden neurons spontaneously. Third, spiking-timing-dependent plasticity (STDP) in memristive synapses is adopted to update weights in situ. Correspondingly, we provide a circuitry example and verify it in LTSPICE. Then the MSNN is benchmarked with the MNIST dataset and analyzed with visualization methods, showing better recognition accuracy (95.15%) than existing SNNs with comparable scales and bio-inspired learning rules. We also consider some non-ideal effects in memristor crossbar array and peripheral circuits. Evaluation results show that the proposed MSNN is robust to finite resolution, circuit noise and writing noise; and larger network scale will help the MSNN alleviate the negative impacts of other non-ideal factors including yield and device-to-device variation. Moreover, the energy efficiency of a MSNN system is estimated to achieve 7:6TOPS/W, showing great potential in low-power applications.