Saved in:
Bibliographic Details
Main Author: Thanh, Cédric Ho
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2501.11919
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909462847029248
author Thanh, Cédric Ho
author_facet Thanh, Cédric Ho
contents The existence of salient semantic clusters in the latent spaces of a neural network during training strongly correlates its final accuracy on classification tasks. This paper proposes a novel fine-tuning method that boosts performance by optimising the formation of these latent clusters, using the Louvain community detection algorithm and a specifically designed clustering loss function. We present preliminary results that demonstrate the viability of this process on classical neural network architectures during fine-tuning on the CIFAR-100 dataset.
format Preprint
id arxiv_https___arxiv_org_abs_2501_11919
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Improving Fine-Tuning with Latent Cluster Correction
Thanh, Cédric Ho
Machine Learning
The existence of salient semantic clusters in the latent spaces of a neural network during training strongly correlates its final accuracy on classification tasks. This paper proposes a novel fine-tuning method that boosts performance by optimising the formation of these latent clusters, using the Louvain community detection algorithm and a specifically designed clustering loss function. We present preliminary results that demonstrate the viability of this process on classical neural network architectures during fine-tuning on the CIFAR-100 dataset.
title Improving Fine-Tuning with Latent Cluster Correction
topic Machine Learning
url https://arxiv.org/abs/2501.11919