ProbLP: A framework for low-precision probabilistic inference

TL;DR

ProbLP is a framework that automates the design of low-precision hardware for probabilistic inference, improving energy efficiency while maintaining accuracy, validated on embedded sensing benchmarks.

Contribution

It introduces an automated approach to optimize low-precision probabilistic inference hardware considering error bounds and energy models.

Findings

Reduces energy consumption in probabilistic inference hardware.

Maintains inference accuracy with low-precision representations.

Successfully validated on embedded sensing benchmarks.

Abstract

Bayesian reasoning is a powerful mechanism for probabilistic inference in smart edge-devices. During such inferences, a low-precision arithmetic representation can enable improved energy efficiency. However, its impact on inference accuracy is not yet understood. Furthermore, general-purpose hardware does not natively support low-precision representation. To address this, we propose ProbLP, a framework that automates the analysis and design of low-precision probabilistic inference hardware. It automatically chooses an appropriate energy-efficient representation based on worst-case error-bounds and hardware energy-models. It generates custom hardware for the resulting inference network exploiting parallelism, pipelining and low-precision operation. The framework is validated on several embedded-sensing benchmarks.