DEBUG-HD: Debugging TinyML models on-device using Hyper-Dimensional computing

TL;DR

DEBUG-HD is an on-device debugging method for TinyML models that uses hyper-dimensional computing to detect input corruptions efficiently, improving reliability in remote environments without cloud access.

Contribution

It introduces a novel HDC encoding technique integrated with neural networks, enabling resource-efficient debugging on tiny devices and outperforming previous methods.

Findings

Achieves 27% better detection of input corruptions on average.

Effective on image and audio datasets.

Resource-efficient for KB-sized TinyML devices.

Abstract

TinyML models often operate in remote, dynamic environments without cloud connectivity, making them prone to failures. Ensuring reliability in such scenarios requires not only detecting model failures but also identifying their root causes. However, transient failures, privacy concerns, and the safety-critical nature of many applications-where systems cannot be interrupted for debugging-complicate the use of raw sensor data for offline analysis. We propose DEBUG-HD, a novel, resource-efficient on-device debugging approach optimized for KB-sized tinyML devices that utilizes hyper-dimensional computing (HDC). Our method introduces a new HDC encoding technique that leverages conventional neural networks, allowing DEBUG-HD to outperform prior binary HDC methods by 27% on average in detecting input corruptions across various image and audio datasets.