Saved in:
Bibliographic Details
Main Authors: Qi, Binchuan, Gong, Wei, Li, Li
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2503.23016
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915218624348160
author Qi, Binchuan
Gong, Wei
Li, Li
author_facet Qi, Binchuan
Gong, Wei
Li, Li
contents In this paper, we adopt a probability distribution estimation perspective to explore the optimization mechanisms of supervised classification using deep neural networks. We demonstrate that, when employing the Fenchel-Young loss, despite the non-convex nature of the fitting error with respect to the model's parameters, global optimal solutions can be approximated by simultaneously minimizing both the gradient norm and the structural error. The former can be controlled through gradient descent algorithms. For the latter, we prove that it can be managed by increasing the number of parameters and ensuring parameter independence, thereby providing theoretical insights into mechanisms such as over-parameterization and random initialization. Ultimately, the paper validates the key conclusions of the proposed method through empirical results, illustrating its practical effectiveness.
format Preprint
id arxiv_https___arxiv_org_abs_2503_23016
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Towards Understanding the Optimization Mechanisms in Deep Learning
Qi, Binchuan
Gong, Wei
Li, Li
Machine Learning
Artificial Intelligence
In this paper, we adopt a probability distribution estimation perspective to explore the optimization mechanisms of supervised classification using deep neural networks. We demonstrate that, when employing the Fenchel-Young loss, despite the non-convex nature of the fitting error with respect to the model's parameters, global optimal solutions can be approximated by simultaneously minimizing both the gradient norm and the structural error. The former can be controlled through gradient descent algorithms. For the latter, we prove that it can be managed by increasing the number of parameters and ensuring parameter independence, thereby providing theoretical insights into mechanisms such as over-parameterization and random initialization. Ultimately, the paper validates the key conclusions of the proposed method through empirical results, illustrating its practical effectiveness.
title Towards Understanding the Optimization Mechanisms in Deep Learning
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2503.23016