Improved post training quantization for heavy tailed model parameters in neural network
Authors:
- Jipeng LiJLJipeng Li
- Xueqiong YuanXYXueqiong Yuan
- Ercan Engin KuruogluEKErcan Engin Kuruoglu
- 7 October 2025
- journal article
- Published by Elsevier in Signal Processing
- Vol. 239
Abstract
No abstract availableFunding Information
- Tsinghua University (ZDSYS20220527171406015)
- Tsinghua Shenzhen International Graduate School (QD2022024C)
- Science, Technology and Innovation Commission of Shenzhen Municipality (JCYJ20220530143002005)
This publication has 9 references indexed in Scilit:
- ClipQ: Clipping Optimization for the Post-Training Quantization of Convolutional Neural NetworkApplied Sciences, 2025
- Notice of Removal: Log-Scale Quantization in Distributed First-Order Methods: Gradient-Based Learning From Distributed DataIEEE Transactions on Automation Science and Engineering, 2025
- Multivariate Myriad Filters Based on Parameter Estimation of Student-$t$ DistributionsSIAM Journal on Imaging Sciences, 2019
- Deep Learning with Low Precision by Half-Wave Gaussian Quantization2017
- Rate-distortion function for -stable sourcesAEU - International Journal of Electronics and Communications, 2016
- Scalar quantisation of heavy-tailed signalsIEE Proceedings - Vision, Image, and Signal Processing, 2000
- Least Lp-norm impulsive noise cancellation with polynomial filtersSignal Processing, 1998
- Least squares quantization in PCMIEEE Transactions on Information Theory, 1982
- An Algorithm for Vector Quantizer DesignIEEE Transactions on Communications, 1980