Making Sigmoid-MSE Great Again: Output Reset Challenges Softmax Cross-Entropy in Neural Network Classification

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Tyagi, Kanishka, Rane, Chinmay, Vaidya, Ketaki, Challgundla, Jeshwanth, Auddy, Soumitro Swapan, Manry, Michael
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913579900338176
author Tyagi, Kanishka
Rane, Chinmay
Vaidya, Ketaki
Challgundla, Jeshwanth
Auddy, Soumitro Swapan
Manry, Michael
author_facet Tyagi, Kanishka
Rane, Chinmay
Vaidya, Ketaki
Challgundla, Jeshwanth
Auddy, Soumitro Swapan
Manry, Michael
contents This study presents a comparative analysis of two objective functions, Mean Squared Error (MSE) and Softmax Cross-Entropy (SCE) for neural network classification tasks. While SCE combined with softmax activation is the conventional choice for transforming network outputs into class probabilities, we explore an alternative approach using MSE with sigmoid activation. We introduce the Output Reset algorithm, which reduces inconsistent errors and enhances classifier robustness. Through extensive experiments on benchmark datasets (MNIST, CIFAR-10, and Fashion-MNIST), we demonstrate that MSE with sigmoid activation achieves comparable accuracy and convergence rates to SCE, while exhibiting superior performance in scenarios with noisy data. Our findings indicate that MSE, despite its traditional association with regression tasks, serves as a viable alternative for classification problems, challenging conventional wisdom about neural network training strategies.
format Preprint
id arxiv_https___arxiv_org_abs_2411_11213
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Making Sigmoid-MSE Great Again: Output Reset Challenges Softmax Cross-Entropy in Neural Network Classification
Tyagi, Kanishka
Rane, Chinmay
Vaidya, Ketaki
Challgundla, Jeshwanth
Auddy, Soumitro Swapan
Manry, Michael
Machine Learning
Artificial Intelligence
This study presents a comparative analysis of two objective functions, Mean Squared Error (MSE) and Softmax Cross-Entropy (SCE) for neural network classification tasks. While SCE combined with softmax activation is the conventional choice for transforming network outputs into class probabilities, we explore an alternative approach using MSE with sigmoid activation. We introduce the Output Reset algorithm, which reduces inconsistent errors and enhances classifier robustness. Through extensive experiments on benchmark datasets (MNIST, CIFAR-10, and Fashion-MNIST), we demonstrate that MSE with sigmoid activation achieves comparable accuracy and convergence rates to SCE, while exhibiting superior performance in scenarios with noisy data. Our findings indicate that MSE, despite its traditional association with regression tasks, serves as a viable alternative for classification problems, challenging conventional wisdom about neural network training strategies.
title Making Sigmoid-MSE Great Again: Output Reset Challenges Softmax Cross-Entropy in Neural Network Classification
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2411.11213