Reference : Impact of Disentanglement on Pruning Neural Networks

	Document type :	Scientific congresses, symposiums and conference proceedings : Unpublished conference
	Discipline(s) :	Physical, chemical, mathematical & earth Sciences : Multidisciplinary, general & others Engineering, computing & technology : Electrical & electronics engineering Engineering, computing & technology : Multidisciplinary, general & others
	Focus Areas :	Computational Sciences; Security, Reliability and Trust
	To cite this reference:	http://hdl.handle.net/10993/56046

Title :	Impact of Disentanglement on Pruning Neural Networks
Language :	English
Author, co-author :	Shneider, Carl [University of Luxembourg > Interdisciplinary Centre for Security, Reliability and Trust (SNT) > CVI2 >]
	Rostami Abendansari, Peyman [University of Luxembourg > Interdisciplinary Centre for Security, Reliability and Trust (SNT) > CVI2 >]
	Kacem, Anis [University of Luxembourg > Interdisciplinary Centre for Security, Reliability and Trust (SNT) > CVI2 >]
	Sinha, Nilotpal [University of Luxembourg > Interdisciplinary Centre for Security, Reliability and Trust (SNT) > CVI2 >]
	Shabayek, Abd El Rahman [University of Luxembourg > Interdisciplinary Centre for Security, Reliability and Trust (SNT) > CVI2 >]
	Aouada, Djamila [University of Luxembourg > Interdisciplinary Centre for Security, Reliability and Trust (SNT) > CVI2 >]
Publication date :	19-Jul-2023
Reviewed work :	Shneider, Carl, Peyman Rostami, Anis Kacem, Nilotpal Sinha, Abd El Rahman Shabayek, and Djamila Aouada. "Impact of Disentanglement on Pruning Neural Networks." arXiv preprint arXiv:2307.09994 (2023).
Peer reviewed :	Yes
Audience :	International
Event name :	International Symposium on Computational Sensing (ISCS23)
Event date :	12-06-2023 to 14-06-2023
Event organizer :	Thomas Feuillen, Amirafshar Moshtaghpour
Event place (city) :	Luxembourg
Event country :	Luxembourg
Keywords :	[en] Deep Learning ; Neural Network Compression ; Variational Autoencoders
Abstract :	[en] Deploying deep learning neural networks on edge devices, to accomplish task specific objectives in the real-world, requires a reduction in their memory footprint, power consumption, and latency. This can be realized via efficient model compression. Disentangled latent representations produced by variational autoencoder (VAE) networks are a promising approach for achieving model compression because they mainly retain task-specific information, discarding useless information for the task at hand. We make use of the Beta-VAE framework combined with a standard criterion for pruning to investigate the impact of forcing the network to learn disentangled representations on the pruning process for the task of classification. In particular, we perform experiments on MNIST and CIFAR10 datasets, examine disentanglement challenges, and propose a path forward for future works.
Research centres :	Interdisciplinary Centre for Security, Reliability and Trust (SnT) > CVI² - Computer Vision Imaging & Machine Intelligence
Funders :	Fonds National de la Recherche - FnR
Name of the research project :	Enabling Learning And Inferring Compact Deep Neural Network Topologies On Edge Devices (ELITE)
Target :	Researchers ; Professionals ; Students
Permalink :	http://hdl.handle.net/10993/56046
DOI :	10.48550/arXiv.2307.09994
Other URL :	https://arxiv.org/abs/2307.09994
FnR project :	FnR ; FNR15965298 > Djamila Aouada > ELITE > Enabling Learning And Inferring Compact Deep Neural Network Topologies On Edge Devices > 01/04/2022 > 31/03/2025 > 2021

	File(s) associated to this reference
	Fulltext file(s): File Commentary Version Size Access Open access Shneider_2307.09994.pdf Author preprint 233.86 kB View/Open

All documents in ORBi^lu are protected by a user license.

 

University of Luxembourg Library |  Feedback |  Legal notices Site Map