Learning to Transfer: Unsupervised Meta Domain Translation
This work addresses the challenge of adapting domain translation models to new domains with limited data, which is incremental as it builds on existing GAN-based methods by incorporating meta-learning.
The paper tackles the problem of unsupervised domain translation by proposing a meta-learning approach to enable models to adapt to new domains with few samples, achieving significant performance improvements over existing methods when each domain has no more than ten training samples.
Unsupervised domain translation has recently achieved impressive performance with Generative Adversarial Network (GAN) and sufficient (unpaired) training data. However, existing domain translation frameworks form in a disposable way where the learning experiences are ignored and the obtained model cannot be adapted to a new coming domain. In this work, we take on unsupervised domain translation problems from a meta-learning perspective. We propose a model called Meta-Translation GAN (MT-GAN) to find good initialization of translation models. In the meta-training procedure, MT-GAN is explicitly trained with a primary translation task and a synthesized dual translation task. A cycle-consistency meta-optimization objective is designed to ensure the generalization ability. We demonstrate effectiveness of our model on ten diverse two-domain translation tasks and multiple face identity translation tasks. We show that our proposed approach significantly outperforms the existing domain translation methods when each domain contains no more than ten training samples.