Saved in:
Bibliographic Details
Main Author: Blázquez, Javier Gil
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.26830
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909000920989696
author Blázquez, Javier Gil
author_facet Blázquez, Javier Gil
contents I propose the \emph{Random Cloud} method, a training-free approach to neural architecture search that discovers minimal feedforward network topologies through stochastic exploration and progressive structural reduction. Unlike post-training pruning methods that require a full train-prune-retrain cycle, this method evaluates randomly initialized networks without backpropagation, progressively reduces their topology, and only trains the best minimal candidate at the end. I evaluate on 7 classification benchmarks against magnitude pruning and random pruning baselines. The Random Cloud matches or outperforms both baselines in 6 of 7 datasets, achieving statistically significant improvements on Sonar ($+4.9$pp accuracy, $p{=}0.017$ vs magnitude pruning) with 87\% parameter reduction. Crucially, the method is faster than both pruning baselines in 4 of 5 datasets (0.67--0.94$\times$ the cost of full training), since it avoids training the full-size network entirely.
format Preprint
id arxiv_https___arxiv_org_abs_2604_26830
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Random Cloud: Finding Minimal Neural Architectures Without Training
Blázquez, Javier Gil
Machine Learning
Artificial Intelligence
I propose the \emph{Random Cloud} method, a training-free approach to neural architecture search that discovers minimal feedforward network topologies through stochastic exploration and progressive structural reduction. Unlike post-training pruning methods that require a full train-prune-retrain cycle, this method evaluates randomly initialized networks without backpropagation, progressively reduces their topology, and only trains the best minimal candidate at the end. I evaluate on 7 classification benchmarks against magnitude pruning and random pruning baselines. The Random Cloud matches or outperforms both baselines in 6 of 7 datasets, achieving statistically significant improvements on Sonar ($+4.9$pp accuracy, $p{=}0.017$ vs magnitude pruning) with 87\% parameter reduction. Crucially, the method is faster than both pruning baselines in 4 of 5 datasets (0.67--0.94$\times$ the cost of full training), since it avoids training the full-size network entirely.
title Random Cloud: Finding Minimal Neural Architectures Without Training
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2604.26830