nanoML for Human Activity Recognition

TL;DR

This paper introduces Differentiable Weightless Neural Networks (DWNs) as a highly energy-efficient and compact solution for Human Activity Recognition on resource-limited devices, demonstrating significant improvements over traditional deep learning models.

Contribution

The paper presents DWNs as a novel nanoML model for HAR, achieving high accuracy with ultra-low energy consumption and memory usage on FPGA hardware.

Findings

Achieved 96.34% and 96.67% accuracy on HAR tasks.

Consumed only 56nJ and 104nJ per sample, respectively.

Up to 926,000x energy savings compared to deep learning methods.

Abstract

Human Activity Recognition (HAR) is critical for applications in healthcare, fitness, and IoT, but deploying accurate models on resource-constrained devices remains challenging due to high energy and memory demands. This paper demonstrates the application of Differentiable Weightless Neural Networks (DWNs) to HAR, achieving competitive accuracies of 96.34% and 96.67% while consuming only 56nJ and 104nJ per sample, with an inference time of just 5ns per sample. The DWNs were implemented and evaluated on an FPGA, showcasing their practical feasibility for energy-efficient hardware deployment. DWNs achieve up to 926,000x energy savings and 260x memory reduction compared to state-of-the-art deep learning methods. These results position DWNs as a nano-machine learning nanoML model for HAR, setting a new benchmark in energy efficiency and compactness for edge and wearable devices, paving the way for ultra-efficient edge AI.