Unsupervised Domain Adaptation using Generative Models and Self-ensembling
arXiv:1812.00479
Abstract
Transferring knowledge across different datasets is an important approach to successfully train deep models with a small-scale target dataset or when few labeled instances are available. In this paper, we aim at developing a model that can generalize across multiple domain shifts, so that this model can adapt from a single source to multiple targets. This can be achieved by randomizing the generation of the data of various styles to mitigate the domain mismatch. First, we present a new adaptation to the CycleGAN model to produce stochastic style transfer between two image batches of different domains. Second, we enhance the classifier performance by using a self-ensembling technique with a teacher and student model to train on both original and generated data. Finally, we present experimental results on three datasets Office-31, Office-Home, and Visual Domain adaptation. The results suggest that selfensembling is better than simple data augmentation with the newly generated data and a single model trained this way can have the best performance across all different transfer tasks.
References in corpus (23)
- Very Deep Convolutional Networks for Large-Scale Image Recognition
- Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
- Learning Transferable Features with Deep Adaptation Networks
- Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results
- Progressive Growing of GANs for Improved Quality, Stability, and Variation
- Fully Convolutional Networks for Semantic Segmentation
- Domain Separation Networks
- VisDA: The Visual Domain Adaptation Challenge
- Adversarial Discriminative Domain Adaptation
- Deep CORAL: Correlation Alignment for Deep Domain Adaptation
- Learning with Pseudo-Ensembles
- Domain Adaptation for Visual Applications: A Comprehensive Survey
- Deep Transfer Learning with Joint Adaptation Networks
- Return of Frustratingly Easy Domain Adaptation
- Mind the Class Weight Bias: Weighted Maximum Mean Discrepancy for Unsupervised Domain Adaptation
- Unsupervised Pixel-Level Domain Adaptation with Generative Adversarial Networks
- Domain Generalization for Object Recognition with Multi-task Autoencoders
- Stacked Generative Adversarial Networks
- Augmented Cyclic Adversarial Learning for Low Resource Domain Adaptation
- Smooth Neighbors on Teacher Graphs for Semi-supervised Learning
- Image to Image Translation for Domain Adaptation
- Deep Adversarial Attention Alignment for Unsupervised Domain Adaptation: the Benefit of Target Expectation Maximization
- Synthetic to Real Adaptation with Generative Correlation Alignment Networks