A Novel Multi-Layer HurNet Architecture: A Fast Alternative to Traditional Deep Neural Networks

Fuente: Zenodo
Saved in:
Bibliographic Details
Main Author: Ben-Hur Varriano
Format: Recurso digital
Published: Zenodo 2026
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866901268801257472
author Ben-Hur Varriano
author_facet Ben-Hur Varriano
contents <p>In this work, we present a detailed mathematical description of a novel multi-layer network architecture called <strong>HurNet</strong>. This model is designed as a lightweight alternative to traditional deep learning frameworks, focusing on computational efficiency and simplified training processes. We derive the mathematical foundations behind the core modules, including input validation, activation function formulation, proximity-based weight selection, and performance metrics. The methodology is supported by a wide array of equations and analytical proofs, and our work is contextualized with references from seminal works in neural computation and optimization. The results suggest that <strong>HurNet</strong> achieves similar accuracy to conventional frameworks while reducing computational time significantly.</p>
format Recurso digital
id zenodo_https___doi_org_10_5281_zenodo_19645833
institution Zenodo
language
publishDate 2026
publisher Zenodo
record_format zenodo
spellingShingle A Novel Multi-Layer HurNet Architecture: A Fast Alternative to Traditional Deep Neural Networks
Ben-Hur Varriano
<p>In this work, we present a detailed mathematical description of a novel multi-layer network architecture called <strong>HurNet</strong>. This model is designed as a lightweight alternative to traditional deep learning frameworks, focusing on computational efficiency and simplified training processes. We derive the mathematical foundations behind the core modules, including input validation, activation function formulation, proximity-based weight selection, and performance metrics. The methodology is supported by a wide array of equations and analytical proofs, and our work is contextualized with references from seminal works in neural computation and optimization. The results suggest that <strong>HurNet</strong> achieves similar accuracy to conventional frameworks while reducing computational time significantly.</p>
title A Novel Multi-Layer HurNet Architecture: A Fast Alternative to Traditional Deep Neural Networks
url https://doi.org/10.5281/zenodo.19645833