查看文章

Few-shot time-series classification aims to learn a classifier on time series data, and the classifier has a fast-adaptive ability that can categorize unseen samples into a class which receives very few labeled training samples. However, conventional time-series classification algorithms fail to tackle the few-shot scenario. Existing fewshot learning methods are proposed to tackle image or text data, and most of them are neuralbased models that lack interpretability. This paper proposes Dual Prototypical Shapelet Networks (DPSN), which not only train a neural networkbased model but also interpret the model from the perspective of representative time series samples and shapelets. In particular, the generated dual prototypical shapelets consist of representative samples that can mostly demonstrate the overall shapes of all samples in the class and discriminative partial-length shapelets that can be used to distinguish different classes. We test DPSN on 22 datasets and show that DPSN outperforms state-of-the-art time-series classification methods, especially when trained with few data.