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

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.

arXiv commentAccepted for publication in Conference on Computer Vision and Pattern Recognition(CVPR), 2018

References in corpus (23)

• Overcoming catastrophic forgetting in neural networks
• Show, Attend and Tell: Neural Image Caption Generation with Visual Attention
• Learning Transferable Features with Deep Adaptation Networks
• Unsupervised Domain Adaptation by Backpropagation
• Unsupervised Domain Adaptation with Residual Transfer Networks
• Fully Convolutional Networks for Semantic Segmentation
• FCNs in the Wild: Pixel-level Adversarial and Constraint-based Adaptation
• Learning multiple visual domains with residual adapters
• Asymmetric Tri-training for Unsupervised Domain Adaptation
• Adversarial Discriminative Domain Adaptation
• A Kernel Method for the Two-Sample Problem
• Few-Shot Adversarial Domain Adaptation
• Continual Learning with Deep Generative Replay
• Return of Frustratingly Easy Domain Adaptation
• Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields
• Mind the Class Weight Bias: Weighted Maximum Mean Discrepancy for Unsupervised Domain Adaptation
• Unsupervised Pixel-Level Domain Adaptation with Generative Adversarial Networks
• Joint Geometrical and Statistical Alignment for Visual Domain Adaptation
• Domain Generalization for Object Recognition with Multi-task Autoencoders
• Associative Domain Adaptation
• When Unsupervised Domain Adaptation Meets Tensor Representations
• Fine-grained Recognition in the Wild: A Multi-Task Domain Adaptation Approach
• Domain Adaptation by Mixture of Alignments of Second- or Higher-Order Scatter Tensors

Cited by in corpus (12)

• Aligning Domain-specific Distribution and Classifier for Cross-domain Classification from Multiple Sources
• Do We Really Need to Access the Source Data? Source Hypothesis Transfer for Unsupervised Domain Adaptation
• Multi-target Unsupervised Domain Adaptation without Exactly Shared Categories
• Domain Generalization Using a Mixture of Multiple Latent Domains
• Domain Attention Consistency for Multi-Source Domain Adaptation
• Mind the Gap: Enlarging the Domain Gap in Open Set Domain Adaptation
• Latent Domain Learning with Dynamic Residual Adapters
• Multi-Source Deep Domain Adaptation for Quality Control in Retail Food Packaging
• Multi-source Domain Adaptation for Visual Sentiment Classification
• Towards Robust Pattern Recognition: A Review
• T-SVDNet: Exploring High-Order Prototypical Correlations for Multi-Source Domain Adaptation
• Dynamic Transfer for Multi-Source Domain Adaptation