@article{Guo_Qiu_Liu_Xue_Zhang_2020, title={Multi-Scale Self-Attention for Text Classification}, volume={34}, url={https://ojs.aaai.org/index.php/AAAI/article/view/6290}, DOI={10.1609/aaai.v34i05.6290}, abstractNote={<p>In this paper, we introduce the prior knowledge, multi-scale structure, into self-attention modules. We propose a Multi-Scale Transformer which uses multi-scale multi-head self-attention to capture features from different scales. Based on the linguistic perspective and the analysis of pre-trained Transformer (BERT) on a huge corpus, we further design a strategy to control the scale distribution for each layer. Results of three different kinds of tasks (21 datasets) show our Multi-Scale Transformer outperforms the standard Transformer consistently and significantly on small and moderate size datasets.</p>}, number={05}, journal={Proceedings of the AAAI Conference on Artificial Intelligence}, author={Guo, Qipeng and Qiu, Xipeng and Liu, Pengfei and Xue, Xiangyang and Zhang, Zheng}, year={2020}, month={Apr.}, pages={7847-7854} }