TY - JOUR
AU - 
AB - Investigating Capsule Networks with Dynamic Routing for Text Classification 1;2 3 1 4 5 6 Wei Zhao , Jianbo Ye , Min Yang , Zeyang Lei , Soufei Zhang , Zhou Zhao Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences 2 3 Tencent Pennsylvania State University Graduate School at Shenzhen, Tsinghua University 5 6 Nanjing University of Posts and Telecommunications Zhejiang University Abstract author’s view in the article is more liberal or more conservative. In this study, we explore capsule networks Earlier efforts in modeling texts have achieved with dynamic routing for text classifica- limited success on text categorization using a sim- tion. We propose three strategies to sta- ple bag-of-words classifier (Joachims, 1998; Mc- bilize the dynamic routing process to al- Callum et al., 1998), implying understanding the leviate the disturbance of some noise cap- meaning of the individual word or n-gram is a sules which may contain “background” in- necessary step towards more sophisticated mod- formation or have not been successfully els. It is therefore not a surprise that distributed trained. A series of experiments are con- representations of words, a.k.a. word embeddings, ducted with capsule networks on six text have received great attention from NLP 
TI - Investigating Capsule Networks with Dynamic Routing for Text Classification
JF - Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing
DO - 10.18653/v1/d18-1350
DA - 2018-01-01
UR - https://www.deepdyve.com/lp/unpaywall/investigating-capsule-networks-with-dynamic-routing-for-text-o4i4e8nSoJ
DP - DeepDyve
ER -