Enregistré dans:
Détails bibliographiques
Auteurs principaux: Shakarami, Ashkan, Yeganeh, Yousef, Farshad, Azade, Nicolè, Lorenzo, Ghidoni, Stefano, Navab, Nassir
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2504.15051
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866911297331789824
author Shakarami, Ashkan
Yeganeh, Yousef
Farshad, Azade
Nicolè, Lorenzo
Ghidoni, Stefano
Navab, Nassir
author_facet Shakarami, Ashkan
Yeganeh, Yousef
Farshad, Azade
Nicolè, Lorenzo
Ghidoni, Stefano
Navab, Nassir
contents Activation functions play a critical role in deep neural networks by shaping gradient flow, optimization stability, and generalization. While ReLU remains widely used due to its simplicity, it suffers from gradient sparsity and dead-neuron issues and offers no adaptivity to input statistics. Smooth alternatives such as Swish and GELU improve gradient propagation but still apply a fixed transformation regardless of the activation distribution. In this paper, we propose VeLU, a Variance-enhanced Learning Unit that introduces variance-aware and distributionally aligned nonlinearity through a principled combination of ArcTan-ArcSin transformations, adaptive scaling, and Wasserstein-2 regularization (Optimal Transport). This design enables VeLU to modulate its response based on local activation variance, mitigate internal covariate shift at the activation level, and improve training stability without adding learnable parameters or architectural overhead. Extensive experiments across six deep neural networks show that VeLU outperforms ReLU, ReLU6, Swish, and GELU on 12 vision benchmarks. The implementation of VeLU is publicly available in GitHub.
format Preprint
id arxiv_https___arxiv_org_abs_2504_15051
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle VeLU: Variance-enhanced Learning Unit for Deep Neural Networks
Shakarami, Ashkan
Yeganeh, Yousef
Farshad, Azade
Nicolè, Lorenzo
Ghidoni, Stefano
Navab, Nassir
Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
Activation functions play a critical role in deep neural networks by shaping gradient flow, optimization stability, and generalization. While ReLU remains widely used due to its simplicity, it suffers from gradient sparsity and dead-neuron issues and offers no adaptivity to input statistics. Smooth alternatives such as Swish and GELU improve gradient propagation but still apply a fixed transformation regardless of the activation distribution. In this paper, we propose VeLU, a Variance-enhanced Learning Unit that introduces variance-aware and distributionally aligned nonlinearity through a principled combination of ArcTan-ArcSin transformations, adaptive scaling, and Wasserstein-2 regularization (Optimal Transport). This design enables VeLU to modulate its response based on local activation variance, mitigate internal covariate shift at the activation level, and improve training stability without adding learnable parameters or architectural overhead. Extensive experiments across six deep neural networks show that VeLU outperforms ReLU, ReLU6, Swish, and GELU on 12 vision benchmarks. The implementation of VeLU is publicly available in GitHub.
title VeLU: Variance-enhanced Learning Unit for Deep Neural Networks
topic Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2504.15051