High-Rate Mixout: Revisiting Mixout for Robust Domain Generalization

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Aminbeidokhti, Masih, Medeiros, Heitor Rapela, Muralidharan, Srikanth, Granger, Eric, Pedersoli, Marco
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911201132281856
author Aminbeidokhti, Masih
Medeiros, Heitor Rapela
Muralidharan, Srikanth
Granger, Eric
Pedersoli, Marco
author_facet Aminbeidokhti, Masih
Medeiros, Heitor Rapela
Muralidharan, Srikanth
Granger, Eric
Pedersoli, Marco
contents Ensembling fine-tuned models initialized from powerful pre-trained weights is a common strategy to improve robustness under distribution shifts, but it comes with substantial computational costs due to the need to train and store multiple models. Dropout offers a lightweight alternative by simulating ensembles through random neuron deactivation; however, when applied to pre-trained models, it tends to over-regularize and disrupt critical representations necessary for generalization. In this work, we investigate Mixout, a stochastic regularization technique that provides an alternative to Dropout for domain generalization. Rather than deactivating neurons, Mixout mitigates overfitting by probabilistically swapping a subset of fine-tuned weights with their pre-trained counterparts during training, thereby maintaining a balance between adaptation and retention of prior knowledge. Our study reveals that achieving strong performance with Mixout on domain generalization benchmarks requires a notably high masking probability of 0.9 for ViTs and 0.8 for ResNets. While this may seem like a simple adjustment, it yields two key advantages for domain generalization: (1) higher masking rates more strongly penalize deviations from the pre-trained parameters, promoting better generalization to unseen domains; and (2) high-rate masking substantially reduces computational overhead, cutting gradient computation by up to 45% and gradient memory usage by up to 90%. Experiments across five domain generalization benchmarks, PACS, VLCS, OfficeHome, TerraIncognita, and DomainNet, using ResNet and ViT architectures, show that our approach, High-rate Mixout, achieves out-of-domain accuracy comparable to ensemble-based methods while significantly reducing training costs.
format Preprint
id arxiv_https___arxiv_org_abs_2510_06955
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle High-Rate Mixout: Revisiting Mixout for Robust Domain Generalization
Aminbeidokhti, Masih
Medeiros, Heitor Rapela
Muralidharan, Srikanth
Granger, Eric
Pedersoli, Marco
Machine Learning
Computer Vision and Pattern Recognition
Ensembling fine-tuned models initialized from powerful pre-trained weights is a common strategy to improve robustness under distribution shifts, but it comes with substantial computational costs due to the need to train and store multiple models. Dropout offers a lightweight alternative by simulating ensembles through random neuron deactivation; however, when applied to pre-trained models, it tends to over-regularize and disrupt critical representations necessary for generalization. In this work, we investigate Mixout, a stochastic regularization technique that provides an alternative to Dropout for domain generalization. Rather than deactivating neurons, Mixout mitigates overfitting by probabilistically swapping a subset of fine-tuned weights with their pre-trained counterparts during training, thereby maintaining a balance between adaptation and retention of prior knowledge. Our study reveals that achieving strong performance with Mixout on domain generalization benchmarks requires a notably high masking probability of 0.9 for ViTs and 0.8 for ResNets. While this may seem like a simple adjustment, it yields two key advantages for domain generalization: (1) higher masking rates more strongly penalize deviations from the pre-trained parameters, promoting better generalization to unseen domains; and (2) high-rate masking substantially reduces computational overhead, cutting gradient computation by up to 45% and gradient memory usage by up to 90%. Experiments across five domain generalization benchmarks, PACS, VLCS, OfficeHome, TerraIncognita, and DomainNet, using ResNet and ViT architectures, show that our approach, High-rate Mixout, achieves out-of-domain accuracy comparable to ensemble-based methods while significantly reducing training costs.
title High-Rate Mixout: Revisiting Mixout for Robust Domain Generalization
topic Machine Learning
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2510.06955