Deep Cocktail Network: Multi-source Unsupervised Domain Adaptation with Category Shift

TL;DR

This paper introduces the Deep Cocktail Network (DCTN), a novel approach for multi-source unsupervised domain adaptation that effectively handles both domain and category shifts among diverse data sources.

Contribution

The paper proposes DCTN, a deep network that models target distribution as a weighted combination of multiple sources and employs adversarial learning with source-specific perplexity scores.

Findings

DCTN outperforms existing methods on three benchmark datasets.

The approach effectively manages domain and category shifts.

Source-specific perplexity scores improve target classification accuracy.

Abstract

Unsupervised domain adaptation (UDA) conventionally assumes labeled source samples coming from a single underlying source distribution. Whereas in practical scenario, labeled data are typically collected from diverse sources. The multiple sources are different not only from the target but also from each other, thus, domain adaptater should not be modeled in the same way. Moreover, those sources may not completely share their categories, which further brings a new transfer challenge called category shift. In this paper, we propose a deep cocktail network (DCTN) to battle the domain and category shifts among multiple sources. Motivated by the theoretical results in \cite{mansour2009domain}, the target distribution can be represented as the weighted combination of source distributions, and, the multi-source unsupervised domain adaptation via DCTN is then performed as two alternating steps: i) It deploys multi-way adversarial learning to minimize the discrepancy between the target and each of the multiple source domains, which also obtains the source-specific perplexity scores to denote the possibilities that a target sample belongs to different source domains. ii) The multi-source category classifiers are integrated with the perplexity scores to classify target sample, and the pseudo-labeled target samples together with source samples are utilized to update the multi-source category classifier and the feature extractor. We evaluate DCTN in three domain adaptation benchmarks, which clearly demonstrate the superiority of our framework.