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

TL;DR

This paper introduces a novel search-based framework for mixed-precision quantization, optimizing bit allocation to enhance performance across various signal processing and machine learning applications.

Contribution

It proposes a new bit allocation framework using PPSO and GC-PSO algorithms, with convergence analysis, applicable to multiple classic fields.

Findings

Superiority of the proposed framework over existing algorithms

Effective bit allocation improves performance in FIR filters, receivers, and gradient descent

Greedy criterion enhances efficiency and feasibility in optimization

Abstract

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.