Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Qian, Yu, Cao, Jian, Li, Xiaoshuang, Zhang, Jie, Li, Hufei, Chen, Jue
Format: Preprint
Veröffentlicht: 2022
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2204.11444
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866916075635998720
author Qian, Yu
Cao, Jian
Li, Xiaoshuang
Zhang, Jie
Li, Hufei
Chen, Jue
author_facet Qian, Yu
Cao, Jian
Li, Xiaoshuang
Zhang, Jie
Li, Hufei
Chen, Jue
contents Structured pruning compresses neural networks by reducing channels (filters) for fast inference and low footprint at run-time. To restore accuracy after pruning, fine-tuning is usually applied to pruned networks. However, too few remaining parameters in pruned networks inevitably bring a great challenge to fine-tuning to restore accuracy. To address this challenge, we propose a novel method that first linearly over-parameterizes the compact layers in pruned networks to enlarge the number of fine-tuning parameters and then re-parameterizes them to the original layers after fine-tuning. Specifically, we equivalently expand the convolution/linear layer with several consecutive convolution/linear layers that do not alter the current output feature maps. Furthermore, we utilize similarity-preserving knowledge distillation that encourages the over-parameterized block to learn the immediate data-to-data similarities of the corresponding dense layer to maintain its feature learning ability. The proposed method is comprehensively evaluated on CIFAR-10 and ImageNet which significantly outperforms the vanilla fine-tuning strategy, especially for large pruning ratio.
format Preprint
id arxiv_https___arxiv_org_abs_2204_11444
institution arXiv
publishDate 2022
record_format arxiv
spellingShingle Boosting Pruned Networks with Linear Over-parameterization
Qian, Yu
Cao, Jian
Li, Xiaoshuang
Zhang, Jie
Li, Hufei
Chen, Jue
Computer Vision and Pattern Recognition
Structured pruning compresses neural networks by reducing channels (filters) for fast inference and low footprint at run-time. To restore accuracy after pruning, fine-tuning is usually applied to pruned networks. However, too few remaining parameters in pruned networks inevitably bring a great challenge to fine-tuning to restore accuracy. To address this challenge, we propose a novel method that first linearly over-parameterizes the compact layers in pruned networks to enlarge the number of fine-tuning parameters and then re-parameterizes them to the original layers after fine-tuning. Specifically, we equivalently expand the convolution/linear layer with several consecutive convolution/linear layers that do not alter the current output feature maps. Furthermore, we utilize similarity-preserving knowledge distillation that encourages the over-parameterized block to learn the immediate data-to-data similarities of the corresponding dense layer to maintain its feature learning ability. The proposed method is comprehensively evaluated on CIFAR-10 and ImageNet which significantly outperforms the vanilla fine-tuning strategy, especially for large pruning ratio.
title Boosting Pruned Networks with Linear Over-parameterization
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2204.11444