Semantically Proportional Patchmix for Few-Shot Learning
This work addresses generalization issues in few-shot learning for image classification, representing an incremental improvement over existing methods.
The paper tackles the problem of insufficient feature generalization in few-shot learning by proposing Semantically Proportional Patchmix (SePPMix), which mixes patches and labels based on semantic information to improve model generalization, achieving effective results on prevalent benchmarks.
Few-shot learning aims to classify unseen classes with only a limited number of labeled data. Recent works have demonstrated that training models with a simple transfer learning strategy can achieve competitive results in few-shot classification. Although excelling at distinguishing training data, these models are not well generalized to unseen data, probably due to insufficient feature representations on evaluation. To tackle this issue, we propose Semantically Proportional Patchmix (SePPMix), in which patches are cut and pasted among training images and the ground truth labels are mixed proportionally to the semantic information of the patches. In this way, we can improve the generalization ability of the model by regional dropout effect without introducing severe label noise. To learn more robust representations of data, we further take rotate transformation on the mixed images and predict rotations as a rule-based regularizer. Extensive experiments on prevalent few-shot benchmarks have shown the effectiveness of our proposed method.