P-Norm Attention Deep CORAL: Extending Correlation Alignment Using Attention and the P-Norm Loss Function

Zhi-Yong Wang,Dae-Ki Kang

doi:10.3390/app11115267

Zhi-Yong Wang, Dae-Ki Kang

Open Access

https://doi.org/10.3390/app11115267

Copy DOI

Abstract

CORrelation ALignment (CORAL) is an unsupervised domain adaptation method that uses a linear transformation to align the covariances of source and target domains. Deep CORAL extends CORAL with a nonlinear transformation using a deep neural network and adds CORAL loss as a part of the total loss to align the covariances of source and target domains. However, there are still two problems to be solved in Deep CORAL: features extracted from AlexNet are not always a good representation of the original data, as well as joint training combined with both the classification and CORAL loss may not be efficient enough to align the distribution of the source and target domain. In this paper, we proposed two strategies: attention to improve the quality of feature maps and the p-norm loss function to align the distribution of the source and target features, further reducing the offset caused by the classification loss function. Experiments on the Office-31 dataset indicate that our proposed methodologies improved Deep CORAL in terms of performance.

Highlights

Introduction[1,2,3], but domain shifts dramatically damage the performance of deep learning methods [4,5]
We defaulted to the labels of all source domain data being given, and the labels of all target data were unknown
Because there are three domains in the Office-31 dataset, we can conduct our experiment on six experiment settings, namely: A → W: (A)mazonas the source domain and (W)ebcam as the target domain; A → D: (A)mazon as the source domain and (D)SLRas the target domain; W → A: (W)ebcam as the source domain and (A)mazon as the target domain; W → D: (W)ebcam as the source domain and (D)SLR

Summary

Introduction

[1,2,3], but domain shifts dramatically damage the performance of deep learning methods [4,5]. In such a scenario, features extracted by a deep neural network, which was pre-trained using existing datasets (called the source domain), can become meaningless for the target task (referred to as the target domain). The different data distributions between the source and target domain will hinder the generalization on the target task, which means the learned knowledge from source domains cannot be transferred to target domains

Methods

Results

Conclusion