Self-ensembling for visual domain adaptation

Abstract

This paper explores the use of self-ensembling for visual domain adaptation problems. Our technique is derived from the mean teacher variant (Tarvainen et al., 2017) of temporal ensembling (Laine et al;, 2017), a technique that achieved state of the art results in the area of semi-supervised learning. We introduce a number of modifications to their approach for challenging domain adaptation scenarios and evaluate its effectiveness. Our approach achieves state of the art results in a variety of benchmarks, including our winning entry in the VISDA-2017 visual domain adaptation challenge. In small image benchmarks, our algorithm not only outperforms prior art, but can also achieve accuracy that is close to that of a classifier trained in a supervised fashion.

arXiv comment20 pages, 3 figure, accepted as a poster at ICLR 2018

References in corpus (12)

• Learning Transferable Features with Deep Adaptation Networks
• Deep Domain Confusion: Maximizing for Domain Invariance
• Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks
• Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results
• Improved Techniques for Training GANs
• Adversarial Discriminative Domain Adaptation
• Asymmetric Tri-training for Unsupervised Domain Adaptation
• Revisiting Batch Normalization For Practical Domain Adaptation
• Virtual Adversarial Training: A Regularization Method for Supervised and Semi-Supervised Learning
• Adversarial Dropout Regularization
• Associative Domain Adaptation
• From source to target and back: symmetric bi-directional adaptive GAN

Cited by in corpus (32)

• A DIRT-T Approach to Unsupervised Domain Adaptation
• An Overview of Deep Semi-Supervised Learning
• Unsupervised Domain Adaptation through Self-Supervision
• Model Adaptation: Historical Contrastive Learning for Unsupervised Domain Adaptation without Source Data
• SemiFL: Semi-Supervised Federated Learning for Unlabeled Clients with Alternate Training
• Rethinking Distributional Matching Based Domain Adaptation
• Self-Training and Adversarial Background Regularization for Unsupervised Domain Adaptive One-Stage Object Detection
• A Prototype-Oriented Framework for Unsupervised Domain Adaptation
• StyleGAN2 Distillation for Feed-forward Image Manipulation
• Unbiased Mean Teacher for Cross-domain Object Detection
• Towards Shape Biased Unsupervised Representation Learning for Domain Generalization
• Network Architecture Search for Domain Adaptation
• Strategy to Increase the Safety of a DNN-based Perception for HAD Systems
• Improving Unsupervised Domain Adaptation with Variational Information Bottleneck
• Neural Snowball for Few-Shot Relation Learning
• Every Node Counts: Self-Ensembling Graph Convolutional Networks for Semi-Supervised Learning
• A Low-Cost Neural ODE with Depthwise Separable Convolution for Edge Domain Adaptation on FPGAs
• Domain Confusion with Self Ensembling for Unsupervised Adaptation
• Explore the Potential Performance of Vision-and-Language Navigation Model: a Snapshot Ensemble Method
• Learning a Domain Classifier Bank for Unsupervised Adaptive Object Detection
• Domain Conditioned Adaptation Network
• Temporal Self-Ensembling Teacher for Semi-Supervised Object Detection
• Consistent Posterior Distributions under Vessel-Mixing: A Regularization for Cross-Domain Retinal Artery/Vein Classification
• Contrastive Learning and Self-Training for Unsupervised Domain Adaptation in Semantic Segmentation
• Learning Classifiers for Domain Adaptation, Zero and Few-Shot Recognition Based on Learning Latent Semantic Parts
• GCCN: Global Context Convolutional Network
• Teacher-Student Consistency For Multi-Source Domain Adaptation
• Attribute-Induced Bias Eliminating for Transductive Zero-Shot Learning
• Online Domain Adaptation for Continuous Cross-Subject Liver Viability Evaluation Based on Irregular Thermal Data
• Teacher-Student Competition for Unsupervised Domain Adaptation
• CFEA: Collaborative Feature Ensembling Adaptation for Domain Adaptation in Unsupervised Optic Disc and Cup Segmentation
• Semi-Supervised Domain Adaptation via Selective Pseudo Labeling and Progressive Self-Training