Challenges and Opportunities in Approximate Bayesian Deep Learning for Intelligent IoT Systems

TL;DR

This paper reviews approximate Bayesian deep learning methods for IoT, discussing their benefits and challenges on edge devices, and explores solutions like pruning and distillation to improve scalability and deployment.

Contribution

It introduces various approximate Bayesian inference techniques tailored for IoT edge devices and discusses strategies to reduce model size and computational demands.

Findings

Approximate Bayesian methods can improve robustness and uncertainty estimation in IoT.

Edge deployment challenges include high computational and storage requirements.

Model pruning and distillation can mitigate resource constraints.

Abstract

Approximate Bayesian deep learning methods hold significant promise for addressing several issues that occur when deploying deep learning components in intelligent systems, including mitigating the occurrence of over-confident errors and providing enhanced robustness to out of distribution examples. However, the computational requirements of existing approximate Bayesian inference methods can make them ill-suited for deployment in intelligent IoT systems that include lower-powered edge devices. In this paper, we present a range of approximate Bayesian inference methods for supervised deep learning and highlight the challenges and opportunities when applying these methods on current edge hardware. We highlight several potential solutions to decreasing model storage requirements and improving computational scalability, including model pruning and distillation methods.