Evaluating Spiking Neural Network On Neuromorphic Platform For Human Activity Recognition

TL;DR

This paper evaluates the energy efficiency and latency of spiking neural networks on neuromorphic hardware for human activity recognition, demonstrating comparable accuracy to traditional neural networks with significantly improved energy-delay performance.

Contribution

It presents a novel evaluation of spiking neural networks on neuromorphic platforms for wearable human activity recognition, highlighting energy and latency benefits.

Findings

Achieved 87.5% accuracy in workout recognition.

Demonstrated two times better energy-delay product compared to traditional neural networks.

Validated the effectiveness of event-based encoding and neuromorphic deployment.

Abstract

Energy efficiency and low latency are crucial requirements for designing wearable AI-empowered human activity recognition systems, due to the hard constraints of battery operations and closed-loop feedback. While neural network models have been extensively compressed to match the stringent edge requirements, spiking neural networks and event-based sensing are recently emerging as promising solutions to further improve performance due to their inherent energy efficiency and capacity to process spatiotemporal data in very low latency. This work aims to evaluate the effectiveness of spiking neural networks on neuromorphic processors in human activity recognition for wearable applications. The case of workout recognition with wrist-worn wearable motion sensors is used as a study. A multi-threshold delta modulation approach is utilized for encoding the input sensor data into spike trains to move the pipeline into the event-based approach. The spikes trains are then fed to a spiking neural network with direct-event training, and the trained model is deployed on the research neuromorphic platform from Intel, Loihi, to evaluate energy and latency efficiency. Test results show that the spike-based workouts recognition system can achieve a comparable accuracy (87.5\%) comparable to the popular milliwatt RISC-V bases multi-core processor GAP8 with a traditional neural network ( 88.1\%) while achieving two times better energy-delay product (0.66 \si{\micro\joule\second} vs. 1.32 \si{\micro\joule\second}).