S-TLLR: STDP-inspired Temporal Local Learning Rule for Spiking Neural Networks

Spiking Neural Networks (SNNs) are biologically plausible models that have been identified as potentially apt for deploying energy-efficient intelligence at the edge, particularly for sequential learning tasks. However, training of SNNs poses significant challenges due to the necessity for precise temporal and spatial credit assignment. Back-propagation through time (BPTT) algorithm, whilst the most widely used method for addressing these issues, incurs a high computational cost due to its temporal dependency. In this work, we propose S-TLLR, a novel threefactor temporal local learning rule inspired by the Spike-Timing Dependent Plasticity (STDP) mechanism, aimed at training deep SNNs on event-based learning tasks. Furthermore, S-TLLR is designed to have low memory and time complexities, which are independent of the number of time steps, rendering it suitable for online learning on low-power edge devices. To demonstrate the scalability of our proposed method, we have conducted extensive evaluations on event-based datasets spanning a wide range of applications, such as image and gesture recognition, audio classification, and optical flow estimation. In all the experiments, S-TLLR achieved high accuracy, comparable to BPTT, with a reduction in memory between 5 50 and multiply-accumulate (MAC) operations between 1.3 6.6 . Over the past decade, the field of artificial intelligence has undergone a remarkable transformation, driven by a prevalent trend of continuously increasing the size and complexity of neural network models. While this approach has yielded remarkable advancements in various cognitive tasks (Brown et al., 2020; Dosovitskiy et al., 2021), it has come at a significant cost: AI systems now demand substantial energy and computational resources. This inherent drawback becomes increasingly apparent when comparing the energy efficiency of current AI systems with the remarkable efficiency exhibited by the human brain (Roy et al., 2019; Gerstner et al., 2014; Christensen et al., 2022; Eshraghian et al., 2023).