Universal Domain Adaptation through Self Supervision

Abstract

Unsupervised domain adaptation methods traditionally assume that all source categories are present in the target domain. In practice, little may be known about the category overlap between the two domains. While some methods address target settings with either partial or open-set categories, they assume that the particular setting is known a priori. We propose a more universally applicable domain adaptation framework that can handle arbitrary category shift, called Domain Adaptative Neighborhood Clustering via Entropy optimization (DANCE). DANCE combines two novel ideas: First, as we cannot fully rely on source categories to learn features discriminative for the target, we propose a novel neighborhood clustering technique to learn the structure of the target domain in a self-supervised way. Second, we use entropy-based feature alignment and rejection to align target features with the source, or reject them as unknown categories based on their entropy. We show through extensive experiments that DANCE outperforms baselines across open-set, open-partial and partial domain adaptation settings. Implementation is available at https://github.com/VisionLearningGroup/DANCE.

Accepted to NeurIPS2020

References in corpus (7)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Distilling the Knowledge in a Neural Network
• Learning Transferable Features with Deep Adaptation Networks
• Deep Domain Confusion: Maximizing for Domain Invariance
• VisDA: The Visual Domain Adaptation Challenge
• Bridging Theory and Algorithm for Domain Adaptation
• Unsupervised Deep Learning by Neighbourhood Discovery

Cited by in corpus (17)

• Multi-Source Unsupervised Domain Adaptation via Pseudo Target Domain
• Universal Domain Adaptation for Remote Sensing Image Scene Classification
• Prototypical Pseudo Label Denoising and Target Structure Learning for Domain Adaptive Semantic Segmentation
• Cross-domain Self-supervised Learning for Domain Adaptation with Few Source Labels
• Self-Supervised Learning for Domain Adaptation on Point-Clouds
• Unsupervised Domain Adaptation in Semantic Segmentation: a Review
• A Prototype-Oriented Framework for Unsupervised Domain Adaptation
• Source Data-absent Unsupervised Domain Adaptation through Hypothesis Transfer and Labeling Transfer
• Adaptive Betweenness Clustering for Semi-Supervised Domain Adaptation
• Self-supervised Learning for Panoptic Segmentation of Multiple Fruit Flower Species
• Prototypical Partial Optimal Transport for Universal Domain Adaptation
• OVANet: One-vs-All Network for Universal Domain Adaptation
• VisDA-2021 Competition Universal Domain Adaptation to Improve Performance on Out-of-Distribution Data
• Domain Adaptation with Auxiliary Target Domain-Oriented Classifier
• Progressively Select and Reject Pseudo-labelled Samples for Open-Set Domain Adaptation
• Uncover and Unlearn Nuisances: Agnostic Fully Test-Time Adaptation
• Universal Multi-Source Domain Adaptation