CoNNect: Connectivity-Based Regularization for Structural Pruning

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Franssen, Christian, Jiang, Jinyang, Peng, Yijie, Heidergott, Bernd
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866915341028818944
author Franssen, Christian
Jiang, Jinyang
Peng, Yijie
Heidergott, Bernd
author_facet Franssen, Christian
Jiang, Jinyang
Peng, Yijie
Heidergott, Bernd
contents Pruning encompasses a range of techniques aimed at increasing the sparsity of neural networks (NNs). These techniques can generally be framed as minimizing a loss function subject to an $L_0$ norm constraint. This paper introduces CoNNect, a novel differentiable regularizer for sparse NN training that ensures connectivity between input and output layers. We prove that CoNNect approximates $L_0$ regularization, guaranteeing maximally connected network structures while avoiding issues like layer collapse. Moreover, CoNNect is easily integrated with established structural pruning strategies. Numerical experiments demonstrate that CoNNect can improve classical pruning strategies and enhance state-of-the-art one-shot pruners, such as DepGraph and LLM-pruner.
format Preprint
id arxiv_https___arxiv_org_abs_2502_00744
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle CoNNect: Connectivity-Based Regularization for Structural Pruning
Franssen, Christian
Jiang, Jinyang
Peng, Yijie
Heidergott, Bernd
Machine Learning
Pruning encompasses a range of techniques aimed at increasing the sparsity of neural networks (NNs). These techniques can generally be framed as minimizing a loss function subject to an $L_0$ norm constraint. This paper introduces CoNNect, a novel differentiable regularizer for sparse NN training that ensures connectivity between input and output layers. We prove that CoNNect approximates $L_0$ regularization, guaranteeing maximally connected network structures while avoiding issues like layer collapse. Moreover, CoNNect is easily integrated with established structural pruning strategies. Numerical experiments demonstrate that CoNNect can improve classical pruning strategies and enhance state-of-the-art one-shot pruners, such as DepGraph and LLM-pruner.
title CoNNect: Connectivity-Based Regularization for Structural Pruning
topic Machine Learning
url https://arxiv.org/abs/2502.00744