Mixed-Precision Quantization: Make the Best Use of Bits Where They Matter Most

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Fang, Yiming, Chen, Li, Chen, Yunfei, Wang, Weidong, You, Changsheng
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911006883577856
author Fang, Yiming
Chen, Li
Chen, Yunfei
Wang, Weidong
You, Changsheng
author_facet Fang, Yiming
Chen, Li
Chen, Yunfei
Wang, Weidong
You, Changsheng
contents Mixed-precision quantization offers superior performance to fixed-precision quantization. It has been widely used in signal processing, communication systems, and machine learning. In mixed-precision quantization, bit allocation is essential. Hence, in this paper, we propose a new bit allocation framework for mixed-precision quantization from a search perspective. First, we formulate a general bit allocation problem for mixed-precision quantization. Then we introduce the penalized particle swarm optimization (PPSO) algorithm to address the integer consumption constraint. To improve efficiency and avoid iterations on infeasible solutions within the PPSO algorithm, a greedy criterion particle swarm optimization (GC-PSO) algorithm is proposed. The corresponding convergence analysis is derived based on dynamical system theory. Furthermore, we apply the above framework to some specific classic fields, i.e., finite impulse response (FIR) filters, receivers, and gradient descent. Numerical examples in each application underscore the superiority of the proposed framework to the existing algorithms.
format Preprint
id arxiv_https___arxiv_org_abs_2412_03101
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Mixed-Precision Quantization: Make the Best Use of Bits Where They Matter Most
Fang, Yiming
Chen, Li
Chen, Yunfei
Wang, Weidong
You, Changsheng
Signal Processing
Information Theory
Mixed-precision quantization offers superior performance to fixed-precision quantization. It has been widely used in signal processing, communication systems, and machine learning. In mixed-precision quantization, bit allocation is essential. Hence, in this paper, we propose a new bit allocation framework for mixed-precision quantization from a search perspective. First, we formulate a general bit allocation problem for mixed-precision quantization. Then we introduce the penalized particle swarm optimization (PPSO) algorithm to address the integer consumption constraint. To improve efficiency and avoid iterations on infeasible solutions within the PPSO algorithm, a greedy criterion particle swarm optimization (GC-PSO) algorithm is proposed. The corresponding convergence analysis is derived based on dynamical system theory. Furthermore, we apply the above framework to some specific classic fields, i.e., finite impulse response (FIR) filters, receivers, and gradient descent. Numerical examples in each application underscore the superiority of the proposed framework to the existing algorithms.
title Mixed-Precision Quantization: Make the Best Use of Bits Where They Matter Most
topic Signal Processing
Information Theory
url https://arxiv.org/abs/2412.03101