Saved in:
Bibliographic Details
Main Authors: Nowdeh, Hossein R., Ji, Jie, Ma, Xiaolong, Afghah, Fatemeh
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2510.24919
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908618502176768
author Nowdeh, Hossein R.
Ji, Jie
Ma, Xiaolong
Afghah, Fatemeh
author_facet Nowdeh, Hossein R.
Ji, Jie
Ma, Xiaolong
Afghah, Fatemeh
contents In multimodal learning, dominant modalities often overshadow others, limiting generalization. We propose Modality-Aware Sharpness-Aware Minimization (M-SAM), a model-agnostic framework that applies to many modalities and supports early and late fusion scenarios. In every iteration, M-SAM in three steps optimizes learning. \textbf{First, it identifies the dominant modality} based on modalities' contribution in the accuracy using Shapley. \textbf{Second, it decomposes the loss landscape}, or in another language, it modulates the loss to prioritize the robustness of the model in favor of the dominant modality, and \textbf{third, M-SAM updates the weights} by backpropagation of modulated gradients. This ensures robust learning for the dominant modality while enhancing contributions from others, allowing the model to explore and exploit complementary features that strengthen overall performance. Extensive experiments on four diverse datasets show that M-SAM outperforms the latest state-of-the-art optimization and gradient manipulation methods and significantly balances and improves multimodal learning.
format Preprint
id arxiv_https___arxiv_org_abs_2510_24919
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Modality-Aware SAM: Sharpness-Aware-Minimization Driven Gradient Modulation for Harmonized Multimodal Learning
Nowdeh, Hossein R.
Ji, Jie
Ma, Xiaolong
Afghah, Fatemeh
Computer Vision and Pattern Recognition
Machine Learning
In multimodal learning, dominant modalities often overshadow others, limiting generalization. We propose Modality-Aware Sharpness-Aware Minimization (M-SAM), a model-agnostic framework that applies to many modalities and supports early and late fusion scenarios. In every iteration, M-SAM in three steps optimizes learning. \textbf{First, it identifies the dominant modality} based on modalities' contribution in the accuracy using Shapley. \textbf{Second, it decomposes the loss landscape}, or in another language, it modulates the loss to prioritize the robustness of the model in favor of the dominant modality, and \textbf{third, M-SAM updates the weights} by backpropagation of modulated gradients. This ensures robust learning for the dominant modality while enhancing contributions from others, allowing the model to explore and exploit complementary features that strengthen overall performance. Extensive experiments on four diverse datasets show that M-SAM outperforms the latest state-of-the-art optimization and gradient manipulation methods and significantly balances and improves multimodal learning.
title Modality-Aware SAM: Sharpness-Aware-Minimization Driven Gradient Modulation for Harmonized Multimodal Learning
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2510.24919