Adaptive Adversarial Cross-Entropy Loss for Sharpness-Aware Minimization

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ratchatorn, Tanapat, Tanaka, Masayuki
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911927681155072
author Ratchatorn, Tanapat
Tanaka, Masayuki
author_facet Ratchatorn, Tanapat
Tanaka, Masayuki
contents Recent advancements in learning algorithms have demonstrated that the sharpness of the loss surface is an effective measure for improving the generalization gap. Building upon this concept, Sharpness-Aware Minimization (SAM) was proposed to enhance model generalization and achieved state-of-the-art performance. SAM consists of two main steps, the weight perturbation step and the weight updating step. However, the perturbation in SAM is determined by only the gradient of the training loss, or cross-entropy loss. As the model approaches a stationary point, this gradient becomes small and oscillates, leading to inconsistent perturbation directions and also has a chance of diminishing the gradient. Our research introduces an innovative approach to further enhancing model generalization. We propose the Adaptive Adversarial Cross-Entropy (AACE) loss function to replace standard cross-entropy loss for SAM's perturbation. AACE loss and its gradient uniquely increase as the model nears convergence, ensuring consistent perturbation direction and addressing the gradient diminishing issue. Additionally, a novel perturbation-generating function utilizing AACE loss without normalization is proposed, enhancing the model's exploratory capabilities in near-optimum stages. Empirical testing confirms the effectiveness of AACE, with experiments demonstrating improved performance in image classification tasks using Wide ResNet and PyramidNet across various datasets. The reproduction code is available online
format Preprint
id arxiv_https___arxiv_org_abs_2406_14329
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Adaptive Adversarial Cross-Entropy Loss for Sharpness-Aware Minimization
Ratchatorn, Tanapat
Tanaka, Masayuki
Machine Learning
Image and Video Processing
Recent advancements in learning algorithms have demonstrated that the sharpness of the loss surface is an effective measure for improving the generalization gap. Building upon this concept, Sharpness-Aware Minimization (SAM) was proposed to enhance model generalization and achieved state-of-the-art performance. SAM consists of two main steps, the weight perturbation step and the weight updating step. However, the perturbation in SAM is determined by only the gradient of the training loss, or cross-entropy loss. As the model approaches a stationary point, this gradient becomes small and oscillates, leading to inconsistent perturbation directions and also has a chance of diminishing the gradient. Our research introduces an innovative approach to further enhancing model generalization. We propose the Adaptive Adversarial Cross-Entropy (AACE) loss function to replace standard cross-entropy loss for SAM's perturbation. AACE loss and its gradient uniquely increase as the model nears convergence, ensuring consistent perturbation direction and addressing the gradient diminishing issue. Additionally, a novel perturbation-generating function utilizing AACE loss without normalization is proposed, enhancing the model's exploratory capabilities in near-optimum stages. Empirical testing confirms the effectiveness of AACE, with experiments demonstrating improved performance in image classification tasks using Wide ResNet and PyramidNet across various datasets. The reproduction code is available online
title Adaptive Adversarial Cross-Entropy Loss for Sharpness-Aware Minimization
topic Machine Learning
Image and Video Processing
url https://arxiv.org/abs/2406.14329