• Home
  • Search
  • Post-training Piecewise Linear Quantization for Deep Neural Networks
  • Open Access IconOpen Access
  • Cite Icon148
  • https://doi.org/10.1007/978-3-030-58536-5_5Copy DOI Icon

Post-training Piecewise Linear Quantization for Deep Neural Networks

  • Jan 1, 2020
  • Jun Fang +5 more
Show More
  • Abstract
  • Literature Map
  • References
  • Citations
  • Similar Papers
Abstract

Quantization plays an important role in the energy-efficient deployment of deep neural networks on resource-limited devices. Post-training quantization is highly desirable since it does not require retraining or access to the full training dataset. The well-established uniform scheme for post-training quantization achieves satisfactory results by converting neural networks from full-precision to 8-bit fixed-point integers. However, it suffers from significant performance degradation when quantizing to lower bit-widths. In this paper, we propose a piecewise linear quantization (PWLQ) scheme (Code will be made available at https://github.com/jun-fang/PWLQ ) to enable accurate approximation for tensor values that have bell-shaped distributions with long tails. Our approach breaks the entire quantization range into non-overlapping regions for each tensor, with each region being assigned an equal number of quantization levels. Optimal breakpoints that divide the entire range are found by minimizing the quantization error. Compared to state-of-the-art post-training quantization methods, experimental results show that our proposed method achieves superior performance on image classification, semantic segmentation, and object detection with minor overhead.

Similar Papers
  • Research Article
  • Citations8

Adversarial Robustness of Deep Convolutional Neural Network-based Image Recognition Models: A Review

  • Aug 28, 2021
  • 雷达学报
  • Hao Sun +4
  • PDF
  • Book Chapter
  • Citations16

QEBVerif: Quantization Error Bound Verification of Neural Networks

  • Jan 01, 2023
  • Yedi Zhang +2
  • Video Transcripts

Towards Low-bit Quantization of Deep Neural Networks with Limited Data

  • Dec 29, 2020
  • Underline Science Inc.
  • Yong Yuan
  • Conference Article
  • Citations11

Lightweight Run-Time Working Memory Compression for Deployment of Deep Neural Networks on Resource-Constrained MCUs

  • Jan 18, 2021
  • Zhepeng Wang +4
  • Conference Article
  • Citations10

Application of shared backbone DNNs in ADAS perception systems

  • Jan 04, 2021
  • Mikhail G Lobanov +1
  • Research Article
  • Citations3

Mobile or FPGA? A Comprehensive Evaluation on Energy Efficiency and a Unified Optimization Framework

  • Sep 30, 2022
  • ACM Transactions on Embedded Computing Systems
  • Geng Yuan +12
  • Conference Article
  • Citations14

QUENN

  • May 08, 2018
  • Miguel De Prado +3
  • Research Article
  • Citations7

Hardware-aware neural architecture search for stochastic computing-based neural networks on tiny devices

  • Dec 22, 2022
  • Journal of Systems Architecture
  • Yuhong Song +6
  • Conference Article
  • Citations5

Architecturally Compressed CNN: An Embedded Realtime Classifier (NXP Bluebox2.0 with RTMaps)

  • Jan 01, 2019
  • Durvesh Pathak +1
  • Research Article
  • Citations1

О сегментации опухолей головного мозга по МРТ-изображениям с применением методов глубокого обучения

  • Jun 16, 2023
  • Journal Of Applied Informatics
  • Eugene Yu Shchetinin
  • Research Article

EffTEE: Efficient Image Classification and Object Detection on Mobile Devices Using Trusted Execution Environments

  • Jan 01, 2025
  • IEEE Access
  • Bin Hu +5
  • Preprint Article

A Countermeasure to Glitch-based Fault Injection Attacks on Deep Neural Networks

  • Sep 04, 2024
  • Preprints.org
  • Seongwoo Hong +4
  • Research Article
  • Citations12

Unsupervised urban scene segmentation via domain adaptation

  • Apr 13, 2020
  • Neurocomputing
  • Lianli Gao +4
  • PDF
  • Research Article
  • Citations1

Advancements in Self-Supervised Learning for Remote Sensing Scene Classification: Present Innovations and Future Outlooks

  • Apr 23, 2024
  • Journal of Artificial Intelligence General science (JAIGS) ISSN:3006-4023
  • José Gabriel Carrasco Ramírez
  • PDF
  • Research Article
  • Citations72

Self-Supervised Learning for Scene Classification in Remote Sensing: Current State of the Art and Perspectives

  • Aug 17, 2022
  • Remote Sensing
  • Paul Berg +2
Cactus Communications logo

Copyright 2026 Cactus Communications. All rights reserved.