Few-Shot Learning as Domain Adaptation: Algorithm and Analysis
This work addresses a key challenge in few-shot learning for AI systems that need to adapt to new classes with limited data, though it is incremental as it builds on existing meta-learning and domain adaptation frameworks.
The paper tackles the problem of poor generalization in few-shot learning due to distribution shift between seen and unseen classes by proposing a domain adaptation prototypical network with attention (DAPNA), which outperforms state-of-the-art methods by significant margins in experiments.
To recognize the unseen classes with only few samples, few-shot learning (FSL) uses prior knowledge learned from the seen classes. A major challenge for FSL is that the distribution of the unseen classes is different from that of those seen, resulting in poor generalization even when a model is meta-trained on the seen classes. This class-difference-caused distribution shift can be considered as a special case of domain shift. In this paper, for the first time, we propose a domain adaptation prototypical network with attention (DAPNA) to explicitly tackle such a domain shift problem in a meta-learning framework. Specifically, armed with a set transformer based attention module, we construct each episode with two sub-episodes without class overlap on the seen classes to simulate the domain shift between the seen and unseen classes. To align the feature distributions of the two sub-episodes with limited training samples, a feature transfer network is employed together with a margin disparity discrepancy (MDD) loss. Importantly, theoretical analysis is provided to give the learning bound of our DAPNA. Extensive experiments show that our DAPNA outperforms the state-of-the-art FSL alternatives, often by significant margins.