On the Robustness of Compressed Models with Class Imbalance

Deep learning (DL) models have been deployed in various platforms, including resource-constrained environments such as edge computing, smartphones, and personal devices. Such deployment requires models to have smaller sizes and memory footprints. To this end, many model compression techniques propos...

Full description

Saved in:
Bibliographic Details
Main Authors: Baraa Saeed Ali, Nabil Sarhan, Mohammed Alawad
Format: Article
Language:English
Published: MDPI AG 2024-11-01
Series:Computers
Subjects:
Online Access:https://www.mdpi.com/2073-431X/13/11/297
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850216708194697216
author Baraa Saeed Ali
Nabil Sarhan
Mohammed Alawad
author_facet Baraa Saeed Ali
Nabil Sarhan
Mohammed Alawad
author_sort Baraa Saeed Ali
collection DOAJ
description Deep learning (DL) models have been deployed in various platforms, including resource-constrained environments such as edge computing, smartphones, and personal devices. Such deployment requires models to have smaller sizes and memory footprints. To this end, many model compression techniques proposed in the literature successfully reduce model sizes and maintain comparable accuracy. However, the robustness of compressed DL models against class imbalance, a natural phenomenon in real-life datasets, is still under-explored. We present a comprehensive experimental study of the performance and robustness of compressed DL models when trained on class-imbalanced datasets. We investigate the robustness of compressed DL models using three popular compression techniques (pruning, quantization, and knowledge distillation) with class-imbalanced variants of the CIFAR-10 dataset and show that compressed DL models are not robust against class imbalance in training datasets. We also show that different compression techniques have varying degrees of impact on the robustness of compressed DL models.
format Article
id doaj-art-43fa7f07f0fd4f339f757e8281bf09ce
institution OA Journals
issn 2073-431X
language English
publishDate 2024-11-01
publisher MDPI AG
record_format Article
series Computers
spelling doaj-art-43fa7f07f0fd4f339f757e8281bf09ce2025-08-20T02:08:14ZengMDPI AGComputers2073-431X2024-11-01131129710.3390/computers13110297On the Robustness of Compressed Models with Class ImbalanceBaraa Saeed Ali0Nabil Sarhan1Mohammed Alawad2Electrical Engineering Department, University of Anbar, Ramadi, Anbar 55431, IraqDepartment of Electrical and Computer Engineering, Wayne State University, Detroit, MI 48202, USADepartment of Electrical and Computer Engineering, Wayne State University, Detroit, MI 48202, USADeep learning (DL) models have been deployed in various platforms, including resource-constrained environments such as edge computing, smartphones, and personal devices. Such deployment requires models to have smaller sizes and memory footprints. To this end, many model compression techniques proposed in the literature successfully reduce model sizes and maintain comparable accuracy. However, the robustness of compressed DL models against class imbalance, a natural phenomenon in real-life datasets, is still under-explored. We present a comprehensive experimental study of the performance and robustness of compressed DL models when trained on class-imbalanced datasets. We investigate the robustness of compressed DL models using three popular compression techniques (pruning, quantization, and knowledge distillation) with class-imbalanced variants of the CIFAR-10 dataset and show that compressed DL models are not robust against class imbalance in training datasets. We also show that different compression techniques have varying degrees of impact on the robustness of compressed DL models.https://www.mdpi.com/2073-431X/13/11/297class imbalancedeep learningmodel compressionrobustness
spellingShingle Baraa Saeed Ali
Nabil Sarhan
Mohammed Alawad
On the Robustness of Compressed Models with Class Imbalance
Computers
class imbalance
deep learning
model compression
robustness
title On the Robustness of Compressed Models with Class Imbalance
title_full On the Robustness of Compressed Models with Class Imbalance
title_fullStr On the Robustness of Compressed Models with Class Imbalance
title_full_unstemmed On the Robustness of Compressed Models with Class Imbalance
title_short On the Robustness of Compressed Models with Class Imbalance
title_sort on the robustness of compressed models with class imbalance
topic class imbalance
deep learning
model compression
robustness
url https://www.mdpi.com/2073-431X/13/11/297
work_keys_str_mv AT baraasaeedali ontherobustnessofcompressedmodelswithclassimbalance
AT nabilsarhan ontherobustnessofcompressedmodelswithclassimbalance
AT mohammedalawad ontherobustnessofcompressedmodelswithclassimbalance