Return of Frustratingly Easy Domain Adaptation

Abstract

Unlike human learning, machine learning often fails to handle changes between training (source) and test (target) input distributions. Such domain shifts, common in practical scenarios, severely damage the performance of conventional machine learning methods. Supervised domain adaptation methods have been proposed for the case when the target data have labels, including some that perform very well despite being "frustratingly easy" to implement. However, in practice, the target domain is often unlabeled, requiring unsupervised adaptation. We propose a simple, effective, and efficient method for unsupervised domain adaptation called CORrelation ALignment (CORAL). CORAL minimizes domain shift by aligning the second-order statistics of source and target distributions, without requiring any target labels. Even though it is extraordinarily simple--it can be implemented in four lines of Matlab code--CORAL performs remarkably well in extensive evaluations on standard benchmark datasets.

Fixed typos. Full paper to appear in AAAI-16. Extended Abstract of the full paper to appear in TASK-CV 2015 workshop

References in corpus (5)

• Learning Transferable Features with Deep Adaptation Networks
• Deep Domain Confusion: Maximizing for Domain Invariance
• Frustratingly Easy Domain Adaptation
• Domain Adaptive Neural Networks for Object Recognition
• Learning from Multiple Outlooks

Cited by in corpus (23)

• Asymmetric Tri-training for Unsupervised Domain Adaptation
• DeepJDOT: Deep Joint Distribution Optimal Transport for Unsupervised Domain Adaptation
• DRIT++: Diverse Image-to-Image Translation via Disentangled Representations
• Maximum Classifier Discrepancy for Unsupervised Domain Adaptation
• Image-Image Domain Adaptation with Preserved Self-Similarity and Domain-Dissimilarity for Person Re-identification
• Deep Cocktail Network: Multi-source Unsupervised Domain Adaptation with Category Shift
• Syn2Real: A New Benchmark forSynthetic-to-Real Visual Domain Adaptation
• SpotTune: Transfer Learning through Adaptive Fine-tuning
• Associative Domain Adaptation
• Unsupervised Neural Domain Adaptation for Document Image Binarization
• Synthetic to Real Adaptation with Generative Correlation Alignment Networks
• From source to target and back: symmetric bi-directional adaptive GAN
• Towards Human Body-Part Learning for Model-Free Gait Recognition
• Domain Alignment with Triplets
• Deep Domain Adaptation by Geodesic Distance Minimization
• Identity Preserving Generative Adversarial Network for Cross-Domain Person Re-identification
• Unsupervised Domain Adaptation using Generative Models and Self-ensembling
• Domain Adaptation for Neural Networks by Parameter Augmentation
• Cross-Domain Sentiment Classification with Contrastive Learning and Mutual Information Maximization
• Affine Transport for Sim-to-Real Domain Adaptation
• Parameter Reference Loss for Unsupervised Domain Adaptation
• Task Guided Compositional Representation Learning for ZDA
• Domain Adaptation for Real-World Single View 3D Reconstruction