NI-UDA: Graph Adversarial Domain Adaptation from Non-shared-and-Imbalanced Big Data to Small Imbalanced Applications

TL;DR

This paper introduces NI-UDA, a novel graph adversarial domain adaptation method that leverages hierarchy graph reasoning and class filtering to improve unsupervised domain adaptation from large, imbalanced datasets to small, imbalanced applications.

Contribution

It proposes Hierarchy Graph Reasoning and Source Classifier Filter to address non-shared and imbalanced class transfer challenges in domain adaptation.

Findings

Significant improvement in F1 scores on benchmark datasets.

Effective handling of sparse and non-shared classes.

Outperforms state-of-the-art adversarial UDA algorithms.

Abstract

We propose a new general Graph Adversarial Domain Adaptation (GADA) based on semantic knowledge reasoning of class structure for solving the problem of unsupervised domain adaptation (UDA) from the big data with non-shared and imbalanced classes to specified small and imbalanced applications (NI-UDA), where non-shared classes mean the label space out of the target domain. Our goal is to leverage priori hierarchy knowledge to enhance domain adversarial aligned feature representation with graph reasoning. In this paper, to address two challenges in NI-UDA, we equip adversarial domain adaptation with Hierarchy Graph Reasoning (HGR) layer and the Source Classifier Filter (SCF). For sparse classes transfer challenge, our HGR layer can aggregate local feature to hierarchy graph nodes by node prediction and enhance domain adversarial aligned feature with hierarchy graph reasoning for sparse classes. Our HGR contributes to learn direct semantic patterns for sparse classes by hierarchy attention in self-attention, non-linear mapping and graph normalization. our SCF is proposed for the challenge of knowledge sharing from non-shared data without negative transfer effect by filtering low-confidence non-shared data in HGR layer. Experiments on two benchmark datasets show our GADA methods consistently improve the state-of-the-art adversarial UDA algorithms, e.g. GADA(HGR) can greatly improve f1 of the MDD by \textbf{7.19\%} and GVB-GD by \textbf{7.89\%} respectively on imbalanced source task in Meal300 dataset. The code is available at https://gadatransfer.wixsite.com/gada.