Unsupervised Domain Adaptation using Generative Models and Self-ensembling
This work addresses the challenge of training deep models with limited labeled data by enabling adaptation from a single source to multiple targets, though it is incremental as it builds on existing methods like CycleGAN and self-ensembling.
The paper tackles the problem of unsupervised domain adaptation across multiple domain shifts by combining a stochastic style transfer method based on CycleGAN with a self-ensembling technique, achieving the best performance across all transfer tasks on datasets like Office-31, Office-Home, and Visual Domain adaptation.
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.