Training Deep Spiking Neural Networks Using Backpropagation

TLDR

Deep spiking neural networks promise lower latency and energy consumption through event‑based computation, yet training them is difficult because spike events are non‑differentiable. This work proposes a novel training technique that treats spiking neuron membrane potentials as differentiable signals, modeling spike discontinuities as noise to enable backpropagation. By allowing standard error backpropagation on spike signals and membrane potentials, the method directly optimizes deep SNNs without indirect conversion. On MNIST and N‑MNIST benchmarks, the technique reduces error on N‑MNIST by more than threefold compared to the best prior SNN, matches conventional CNN accuracy, and achieves equivalent performance with roughly five times fewer computational operations.

Abstract

Deep spiking neural networks (SNNs) hold the potential for improving the latency and energy efficiency of deep neural networks through data-driven event-based computation. However, training such networks is difficult due to the non-differentiable nature of spike events. In this paper, we introduce a novel technique, which treats the membrane potentials of spiking neurons as differentiable signals, where discontinuities at spike times are considered as noise. This enables an error backpropagation mechanism for deep SNNs that follows the same principles as in conventional deep networks, but works directly on spike signals and membrane potentials. Compared with previous methods relying on indirect training and conversion, our technique has the potential to capture the statistics of spikes more precisely. We evaluate the proposed framework on artificially generated events from the original MNIST handwritten digit benchmark, and also on the N-MNIST benchmark recorded with an event-based dynamic vision sensor, in which the proposed method reduces the error rate by a factor of more than three compared to the best previous SNN, and also achieves a higher accuracy than a conventional convolutional neural network (CNN) trained and tested on the same data. We demonstrate in the context of the MNIST task that thanks to their event-driven operation, deep SNNs (both fully connected and convolutional) trained with our method achieve accuracy equivalent with conventional neural networks. In the N-MNIST example, equivalent accuracy is achieved with about five times fewer computational operations.