Generalized Adaptation for Few-Shot Learning

Abstract

Many Few-Shot Learning research works have two stages: pre-training base model and adapting to novel model. In this paper, we propose to use closed-form base learner, which constrains the adapting stage with pre-trained base model to get better generalized novel model. Following theoretical analysis proves its rationality as well as indication of how to train a well-generalized base model. We then conduct experiments on four benchmarks and achieve state-of-the-art performance in all cases. Notably, we achieve the accuracy of 87.75% on 5-shot miniImageNet which approximately outperforms existing methods by 10%.

We found a bug in the script in reexamining some of this work. We decide to withdraw for further modification

References in corpus (9)

• Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
• Prototypical Networks for Few-shot Learning
• Wide Residual Networks
• Matching Networks for One Shot Learning
• Meta-SGD: Learning to Learn Quickly for Few-Shot Learning
• Meta-Learning with Latent Embedding Optimization
• Meta-learning with differentiable closed-form solvers
• Few-Shot Learning with Metric-Agnostic Conditional Embeddings
• Dynamic Few-Shot Visual Learning without Forgetting

Cited by in corpus (3)

• Leveraging the Feature Distribution in Transfer-based Few-Shot Learning
• Unraveling Meta-Learning: Understanding Feature Representations for Few-Shot Tasks
• Graph-based Interpolation of Feature Vectors for Accurate Few-Shot Classification