A comparative study of neural network pruning strategies for industrial applications
IntroductionIn recent years, Deep Learning (DL) and Artificial Neural Networks (ANNs) have transformed industrial applications by providing automation in complex tasks such as anomaly detection and predictive maintenance. However, traditional DL models often need significant computational resources,...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2025-04-01
|
| Series: | Frontiers in Computer Science |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fcomp.2025.1563942/full |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850182921011331072 |
|---|---|
| author | Amirhossein Douzandeh Zenoozi Laura Erhan Antonio Liotta Lucia Cavallaro |
| author_facet | Amirhossein Douzandeh Zenoozi Laura Erhan Antonio Liotta Lucia Cavallaro |
| author_sort | Amirhossein Douzandeh Zenoozi |
| collection | DOAJ |
| description | IntroductionIn recent years, Deep Learning (DL) and Artificial Neural Networks (ANNs) have transformed industrial applications by providing automation in complex tasks such as anomaly detection and predictive maintenance. However, traditional DL models often need significant computational resources, making them unsuitable for resource-constrained edge devices. This paper explores the potential of sparse ANNs to address these challenges, focusing on their application in industrial settings.MethodsWe perform an experimental comparison of pruning techniques, including the Pre-Training, In-Training, Post-Training, and SET method, applied to the VGG16 and ResNet18 architectures, and conduct a systematic analysis of pruning methodologies alongside the effects of varying sparsity levels, to analyze their efficiency in anomaly detection and object classification tasks. Key metrics such as training accuracy, inference time, and energy consumption are analyzed to assess the feasibility of deploying sparse models on edge devices.Results and discussionOur results demonstrate that sparse ANNs, particularly when pruned using the SET method, achieve energy savings without compromising accuracy, making them suitable for industrial applications. This work highlights the potential of sparse neural networks to boost sustainability and efficiency in industrial environments, paving the way for their large adoption in edge computing scenarios. |
| format | Article |
| id | doaj-art-288019c4eaa846d3b343d86f8f528ce5 |
| institution | OA Journals |
| issn | 2624-9898 |
| language | English |
| publishDate | 2025-04-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Computer Science |
| spelling | doaj-art-288019c4eaa846d3b343d86f8f528ce52025-08-20T02:17:29ZengFrontiers Media S.A.Frontiers in Computer Science2624-98982025-04-01710.3389/fcomp.2025.15639421563942A comparative study of neural network pruning strategies for industrial applicationsAmirhossein Douzandeh Zenoozi0Laura Erhan1Antonio Liotta2Lucia Cavallaro3Department of Engineering, Free University of Bozen-Bolzano, Bolzano, ItalyDepartment of Engineering, Free University of Bozen-Bolzano, Bolzano, ItalyDepartment of Engineering, Free University of Bozen-Bolzano, Bolzano, ItalyData Science Department, Institute for Computing and Information Sciences, Radboud University, Nijmegen, NetherlandsIntroductionIn recent years, Deep Learning (DL) and Artificial Neural Networks (ANNs) have transformed industrial applications by providing automation in complex tasks such as anomaly detection and predictive maintenance. However, traditional DL models often need significant computational resources, making them unsuitable for resource-constrained edge devices. This paper explores the potential of sparse ANNs to address these challenges, focusing on their application in industrial settings.MethodsWe perform an experimental comparison of pruning techniques, including the Pre-Training, In-Training, Post-Training, and SET method, applied to the VGG16 and ResNet18 architectures, and conduct a systematic analysis of pruning methodologies alongside the effects of varying sparsity levels, to analyze their efficiency in anomaly detection and object classification tasks. Key metrics such as training accuracy, inference time, and energy consumption are analyzed to assess the feasibility of deploying sparse models on edge devices.Results and discussionOur results demonstrate that sparse ANNs, particularly when pruned using the SET method, achieve energy savings without compromising accuracy, making them suitable for industrial applications. This work highlights the potential of sparse neural networks to boost sustainability and efficiency in industrial environments, paving the way for their large adoption in edge computing scenarios.https://www.frontiersin.org/articles/10.3389/fcomp.2025.1563942/fullsparse neural networkspruning techniquesindustrial applicationsenergy efficiencyedge devices |
| spellingShingle | Amirhossein Douzandeh Zenoozi Laura Erhan Antonio Liotta Lucia Cavallaro A comparative study of neural network pruning strategies for industrial applications Frontiers in Computer Science sparse neural networks pruning techniques industrial applications energy efficiency edge devices |
| title | A comparative study of neural network pruning strategies for industrial applications |
| title_full | A comparative study of neural network pruning strategies for industrial applications |
| title_fullStr | A comparative study of neural network pruning strategies for industrial applications |
| title_full_unstemmed | A comparative study of neural network pruning strategies for industrial applications |
| title_short | A comparative study of neural network pruning strategies for industrial applications |
| title_sort | comparative study of neural network pruning strategies for industrial applications |
| topic | sparse neural networks pruning techniques industrial applications energy efficiency edge devices |
| url | https://www.frontiersin.org/articles/10.3389/fcomp.2025.1563942/full |
| work_keys_str_mv | AT amirhosseindouzandehzenoozi acomparativestudyofneuralnetworkpruningstrategiesforindustrialapplications AT lauraerhan acomparativestudyofneuralnetworkpruningstrategiesforindustrialapplications AT antonioliotta acomparativestudyofneuralnetworkpruningstrategiesforindustrialapplications AT luciacavallaro acomparativestudyofneuralnetworkpruningstrategiesforindustrialapplications AT amirhosseindouzandehzenoozi comparativestudyofneuralnetworkpruningstrategiesforindustrialapplications AT lauraerhan comparativestudyofneuralnetworkpruningstrategiesforindustrialapplications AT antonioliotta comparativestudyofneuralnetworkpruningstrategiesforindustrialapplications AT luciacavallaro comparativestudyofneuralnetworkpruningstrategiesforindustrialapplications |