Mixed Set Domain Adaptation

TL;DR

This paper introduces Mixed Set Domain Adaptation (MSDA), addressing scenarios where source categories come from different domains, and proposes a feature element-wise weighting method to reduce distribution discrepancies, improving adaptation performance.

Contribution

The paper defines the MSDA problem and proposes a novel feature element-wise weighting method to handle intra-source domain discrepancies.

Findings

MSDA is a significant extension of traditional domain adaptation.

FEW method effectively reduces distribution discrepancy.

Experimental results demonstrate improved adaptation performance.

Abstract

In the settings of conventional domain adaptation, categories of the source dataset are from the same domain (or domains for multi-source domain adaptation), which is not always true in reality. In this paper, we propose \textbf{\textit{Mixed Set Domain Adaptation} (MSDA)}. Under the settings of MSDA, different categories of the source dataset are not all collected from the same domain(s). For instance, category $1\sim k$ are collected from domain $\alpha$ while category $k+1\sim c$ are collected from domain $\beta$. Under such situation, domain adaptation performance will be further influenced because of the distribution discrepancy inside the source data. A feature element-wise weighting (FEW) method that can reduce distribution discrepancy between different categories is also proposed for MSDA. Experimental results and quality analysis show the significance of solving MSDA problem and the effectiveness of the proposed method.