EdgeSpike让边缘设备用脉冲神经网络实现低功耗自主感知。
EdgeSpike: Spiking Neural Networks for Low-Power Autonomous Sensing in Edge IoT Architectures
- 混合训练+硬件感知搜索，适配多种低功耗芯片
- 推理能耗降18~47倍，精度仅差1.2个百分点
- 支持设备端持续自适应，适合长期部署的物联网场景
我们提出EdgeSpike，一种面向边缘物联网架构中自主低功耗感知的脉冲神经网络（SNN）协同设计框架。该框架统一了（i）混合代理梯度与直接编码的训练流程，（ii）基于每推理能耗与内存预算约束的硬件感知神经架构搜索（NAS），（iii）面向Intel Loihi 2、SpiNNaker 2及通用ARM Cortex-M微控制器的事件驱动运行时，配备定制的稀疏脉冲SIMD内核，以及（iv）轻量级局部可塑性规则，实现无需反向传播的持续设备端自适应。在三个硬件平台上的五个传感任务（关键词识别、振动故障检测、表面肌电手势识别、77 GHz雷达人体活动分类、结构健康声发射监测）中评估，EdgeSpike平均分类准确率达91.4%，较强的INT8卷积神经网络基线（平均92.6%）仅低1.2个百分点；在神经形态硬件上能将单次推理能耗降低18至47倍（平均31倍），在Cortex-M上降低4.6至7.9倍（平均6.1倍）。所有15种任务-硬件组合的端到端延迟均≤9.4毫秒。一个为期七个月、包含64个节点的无线实地部署验证了电池寿命延长6.3倍（从312天增至1978天，每节点2瓦时），且在设备端自适应下季节漂移导致的精度下降仅为0.7百分点，无自适应时为2.1百分点。硬件感知NAS共评估8400个候选模型，获得12点帕累托前沿。EdgeSpike将开源，提供可复现的训练流程、硬件可移植运行时及基准测试套件。
原文摘要 · Abstract (English)
We propose EdgeSpike, a co-designed spiking neural network (SNN) framework for autonomous low-power sensing in edge Internet of Things (IoT) architectures. EdgeSpike unifies (i) a hybrid surrogate-gradient and direct-encoding training pipeline, (ii) a hardware-aware neural architecture search (NAS) bounded by per-inference energy and memory budgets, (iii) an event-driven runtime targeting Intel Loihi 2, SpiNNaker 2, and commodity ARM Cortex-M microcontrollers with custom spike-sparse SIMD kernels, and (iv) a lightweight local plasticity rule enabling continual on-device adaptation without backpropagation. The framework is evaluated across five sensing tasks (keyword spotting, vibration-based machine fault detection, surface electromyography gesture recognition, 77 GHz radar human-activity classification, and structural-health acoustic-emission monitoring) on three hardware targets. EdgeSpike achieves a mean classification accuracy of 91.4%, within 1.2 percentage points (pp) of strong INT8 convolutional neural network (CNN) baselines (mean 92.6%), while reducing energy per inference by 18x to 47x on neuromorphic hardware (mean 31x) and by 4.6x to 7.9x on Cortex-M (mean 6.1x). End-to-end latency remains at or below 9.4 ms across all 15 task-hardware configurations. A seven-month, 64-node wireless field deployment confirms a 6.3x extension in projected battery lifetime (from 312 to 1978 days at 2 Wh per node) and bounded accuracy degradation under seasonal drift (0.7 pp with on-device adaptation versus 2.1 pp without). Hardware-aware NAS evaluates 8400 candidates and yields a 12-point Pareto front. EdgeSpike will be released as open source with reproducible training pipelines, hardware-portable runtimes, and benchmark suites.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。