Partial Transfer Learning with Selective Adversarial Networks

Abstract

Adversarial learning has been successfully embedded into deep networks to learn transferable features, which reduce distribution discrepancy between the source and target domains. Existing domain adversarial networks assume fully shared label space across domains. In the presence of big data, there is strong motivation of transferring both classification and representation models from existing big domains to unknown small domains. This paper introduces partial transfer learning, which relaxes the shared label space assumption to that the target label space is only a subspace of the source label space. Previous methods typically match the whole source domain to the target domain, which are prone to negative transfer for the partial transfer problem. We present Selective Adversarial Network (SAN), which simultaneously circumvents negative transfer by selecting out the outlier source classes and promotes positive transfer by maximally matching the data distributions in the shared label space. Experiments demonstrate that our models exceed state-of-the-art results for partial transfer learning tasks on several benchmark datasets.

References in corpus (4)

• Natural Language Processing (almost) from Scratch
• DeCAF: A Deep Convolutional Activation Feature for Generic Visual Recognition
• Deep Domain Confusion: Maximizing for Domain Invariance
• Domain Adaptation: Learning Bounds and Algorithms

Cited by in corpus (13)

• Deep Visual Domain Adaptation: A Survey
• Racial Faces in-the-Wild: Reducing Racial Bias by Information Maximization Adaptation Network
• Importance Weighted Adversarial Nets for Partial Domain Adaptation
• Coupling Adversarial Learning with Selective Voting Strategy for Distribution Alignment in Partial Domain Adaptation
• Prototypical Partial Optimal Transport for Universal Domain Adaptation
• On Universal Black-Box Domain Adaptation
• Class-imbalanced Domain Adaptation: An Empirical Odyssey
• One Self-Configurable Model to Solve Many Abstract Visual Reasoning Problems
• Domain Adaptation Meets Disentangled Representation Learning and Style Transfer
• Relation Adversarial Network for Low Resource Knowledge Graph Completion
• Cost-Effective Training of Deep CNNs with Active Model Adaptation
• Unsupervised Open Domain Recognition by Semantic Discrepancy Minimization
• Partially-Shared Variational Auto-encoders for Unsupervised Domain Adaptation with Target Shift