N-ReLU: Zero-Mean Stochastic Extension of ReLU

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Manik, Md Motaleb Hossen, Islam, Md Zabirul, Wang, Ge
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866908641651589120
author Manik, Md Motaleb Hossen
Islam, Md Zabirul
Wang, Ge
author_facet Manik, Md Motaleb Hossen
Islam, Md Zabirul
Wang, Ge
contents Activation functions are fundamental for enabling nonlinear representations in deep neural networks. However, the standard rectified linear unit (ReLU) often suffers from inactive or "dead" neurons caused by its hard zero cutoff. To address this issue, we introduce N-ReLU (Noise-ReLU), a zero-mean stochastic extension of ReLU that replaces negative activations with Gaussian noise while preserving the same expected output. This expectation-aligned formulation maintains gradient flow in inactive regions and acts as an annealing-style regularizer during training. Experiments on the MNIST dataset using both multilayer perceptron (MLP) and convolutional neural network (CNN) architectures show that N-ReLU achieves accuracy comparable to or slightly exceeding that of ReLU, LeakyReLU, PReLU, GELU, and RReLU at moderate noise levels (sigma = 0.05-0.10), with stable convergence and no dead neurons observed. These results demonstrate that lightweight Gaussian noise injection offers a simple yet effective mechanism to enhance optimization robustness without modifying network structures or introducing additional parameters.
format Preprint
id arxiv_https___arxiv_org_abs_2511_07559
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle N-ReLU: Zero-Mean Stochastic Extension of ReLU
Manik, Md Motaleb Hossen
Islam, Md Zabirul
Wang, Ge
Machine Learning
Artificial Intelligence
Activation functions are fundamental for enabling nonlinear representations in deep neural networks. However, the standard rectified linear unit (ReLU) often suffers from inactive or "dead" neurons caused by its hard zero cutoff. To address this issue, we introduce N-ReLU (Noise-ReLU), a zero-mean stochastic extension of ReLU that replaces negative activations with Gaussian noise while preserving the same expected output. This expectation-aligned formulation maintains gradient flow in inactive regions and acts as an annealing-style regularizer during training. Experiments on the MNIST dataset using both multilayer perceptron (MLP) and convolutional neural network (CNN) architectures show that N-ReLU achieves accuracy comparable to or slightly exceeding that of ReLU, LeakyReLU, PReLU, GELU, and RReLU at moderate noise levels (sigma = 0.05-0.10), with stable convergence and no dead neurons observed. These results demonstrate that lightweight Gaussian noise injection offers a simple yet effective mechanism to enhance optimization robustness without modifying network structures or introducing additional parameters.
title N-ReLU: Zero-Mean Stochastic Extension of ReLU
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2511.07559