SpikON: A Dual-Parallel and Efficient Accelerator for Online Spiking Neural Networks Learning

2026-06-29

AI Summary

This work addresses the challenges of low training accuracy, poor scalability, and insufficient hardware efficiency in supervised learning for online spiking neural networks (SNNs). To overcome these limitations, the authors propose SpikON, the first algorithm-hardware co-design framework enabling end-to-end online supervised SNN learning. SpikON introduces several key innovations: learnable firing thresholds, temporal scaling with weight centering, and a cascaded computation reuse mechanism. It also features a dual-parallel engine and a customized SIMD core to support concurrent forward and backward execution. Experimental results demonstrate that SpikON reduces training latency by 32.2% and energy consumption by 35.0% compared to baseline approaches. Moreover, it achieves 7.2–11.5× higher training throughput and 15.8–26.8× better energy efficiency than Apple M4 GPU and TPU-like accelerators.

Abstract

Spiking neural networks (SNNs) have emerged as a promising paradigm for energy-efficient brain-inspired computing. However, existing online unsupervised SNN learning suffers from low training accuracy and poor scalability. Although current online supervised learning algorithms perform well on large-scale datasets and networks, the non-hardware-friendly operations hinder efficient edge deployment. In this work, we propose SpikON, the first algorithm-hardware co-design framework for efficient and scalable end-to-end online supervised SNN learning. We first propose the learnable threshold through time and scaled weight centralization through time techniques to address the inefficiency of traditional algorithms. Moreover, to reduce latency and energy consumption, we introduce the novel training dataflow and cascade computation reuse scheme for SNNs that allows concurrent forward-backward computation and temporal reuse across timesteps. We further design the dedicated SNN accelerator with a dual-parallel engine and customized SIMD-based SNN core for efficient end-to-end online learning. Experiments show that the SpikON algorithm achieves 32.2% and 35.0% reductions in training latency and energy consumption over the baseline, without sacrificing accuracy. Moreover, the SpikON co-design achieves 7.2x (11.5x) and 26.8x (15.8x) training throughput (energy efficiency) compared with the edge Apple M4 GPU and TPU-like accelerator, respectively. The code is available at https://github.com/peilin-chen/SpikON.

Problem

• Spiking Neural Networks
• Online Learning
• Hardware Efficiency
• Edge Deployment
• Scalability

Innovation

• online supervised learning
• algorithm-hardware co-design
• spiking neural networks
• dual-parallel accelerator
• computation reuse