域泛化问题中基于Mixup与对比损失的特征转换模型
A feature transfer model with Mixup and contrastive loss in domain generalization
王岳松 1张洪1
作者信息
- 1. 中国科学技术大学管理学院统计与金融系,安徽合肥 230026
- 折叠
摘要
当表示基础数据分布的域在训练和测试数据集之间存在差异时,传统的深度神经网络的性能会大幅下降.域泛化方法旨在仅使用源域的训练数据来提高在未知目标域上的泛化能力.主流的域泛化算法通常对一些流行的特征提取网络(如ResNet)进行修改,或者在特征提取网络之后添加更复杂的参数模块.流行的特征提取网络通常在大规模数据集上进行了较好的预训练,因此具有较强的特征提取能力,而对其进行修改会削弱这种能力.添加更复杂的参数模块会导致更深的网络,并且对计算资源要求更高.本文基于域泛化中流行的特征提取网络,提出了一种新的特征转换模型,不做任何更改或添加任何模块.通过结合对比损失和数据增强策略(即Mixup),该特征转换模型的泛化能力得到了提升,并提出了一种新的样本选择策略来与Mixup和对比损失相协作.在基准数据集PACS和Domainnet上的实验结果表明,该方法优于传统的域泛化方法.
Abstract
When domains,which represent underlying data distributions,differ between training and test datasets,tradi-tional deep neural networks suffer from a substantial drop in their performance.Domain generalization methods aim to boost generalizability on an unseen target domain by using only training data from source domains.Mainstream domain generalization algorithms usually make modifications on some popular feature extraction networks such as ResNet,or add more complex parameter modules after the feature extraction networks.Popular feature extraction networks are usually well pre-trained on large-scale datasets,so they have strong feature extraction abilities,while modifications can weaken such abilities.Adding more complex parameter modules results in a deeper network and is much more computationally de-manding.In this paper,we propose a novel feature transfer model based on popular feature extraction networks in domain generalization,without making any changes or adding any module.The generalizability of this feature transfer model is boosted by incorporating a contrastive loss and a data augmentation strategy(i.e.,Mixup),and a new sample selection strategy is proposed to coordinate Mixup and contrastive loss.Experiments on the benchmarks PACS and Domainnet demonstrate the superiority of our proposed method against conventional domain generalization methods.
关键词
对比损失/数据增强/深度神经网络/域泛化/特征转换Key words
contrastive loss/data augmentation/deep neural network/domain generalization/feature transfer引用本文复制引用
基金项目
国家自然科学基金(12171451)
Anhui Center for Applied Mathematics()
出版年
2024
NETL
NSTL
万方数据