{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/text-classification-improved-by-integrating","title":"Text Classification Improved by Integrating Bidirectional LSTM with Two-dimensional Max Pooling","arxiv_id":"1611.06639","date":"2016-11-21","proceeding":"COLING 2016 12","authors":["Peng Zhou","Zhenyu Qi","Suncong Zheng","Jiaming Xu","Hongyun Bao","Bo Xu"],"abstract":"Recurrent Neural Network (RNN) is one of the most popular architectures used\nin Natural Language Processsing (NLP) tasks because its recurrent structure is\nvery suitable to process variable-length text. RNN can utilize distributed\nrepresentations of words by first converting the tokens comprising each text\ninto vectors, which form a matrix. And this matrix includes two dimensions: the\ntime-step dimension and the feature vector dimension. Then most existing models\nusually utilize one-dimensional (1D) max pooling operation or attention-based\noperation only on the time-step dimension to obtain a fixed-length vector.\nHowever, the features on the feature vector dimension are not mutually\nindependent, and simply applying 1D pooling operation over the time-step\ndimension independently may destroy the structure of the feature\nrepresentation. On the other hand, applying two-dimensional (2D) pooling\noperation over the two dimensions may sample more meaningful features for\nsequence modeling tasks. To integrate the features on both dimensions of the\nmatrix, this paper explores applying 2D max pooling operation to obtain a\nfixed-length representation of the text. This paper also utilizes 2D\nconvolution to sample more meaningful information of the matrix. Experiments\nare conducted on six text classification tasks, including sentiment analysis,\nquestion classification, subjectivity classification and newsgroup\nclassification. Compared with the state-of-the-art models, the proposed models\nachieve excellent performance on 4 out of 6 tasks. Specifically, one of the\nproposed models achieves highest accuracy on Stanford Sentiment Treebank binary\nclassification and fine-grained classification tasks.","url_abs":"http://arxiv.org/abs/1611.06639v1","url_pdf":"http://arxiv.org/pdf/1611.06639v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"text-classification-improved-by-integrating","repo_url":"https://github.com/ManuelVs/NNForTextClassification","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"text-classification-improved-by-integrating","repo_url":"https://github.com/ManuelVs/NeuralNetworks","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"text-classification-improved-by-integrating","repo_url":"https://github.com/sinhasam/Sentalyze","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"binary-classification","task_name":"Binary Classification"},{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"sentiment-analysis","task_name":"Sentiment Analysis"},{"task_slug":"text-classification","task_name":"Text Classification"}],"methods":[{"method_slug":"max-pooling","method_name":"Max Pooling"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/sentiment-analysis-on-sst-2-binary","task":"Sentiment Analysis","dataset":"SST-2 Binary classification","model":"BLSTM-2DCNN","rank_in_archive_order":64,"of":87,"metrics":{"Accuracy":"89.5"},"uses_additional_data":false},{"leaderboard":"/sota/text-classification-on-trec-6","task":"Text Classification","dataset":"TREC-6","model":"LSTM-CNN","rank_in_archive_order":6,"of":19,"metrics":{"Error":"3.9"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1611.06639","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}