{"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/semi-supervised-sequence-modeling-with-cross","title":"Semi-Supervised Sequence Modeling with Cross-View Training","arxiv_id":"1809.08370","date":"2018-09-22","proceeding":"EMNLP 2018 10","authors":["Kevin Clark","Minh-Thang Luong","Christopher D. Manning","Quoc V. Le"],"abstract":"Unsupervised representation learning algorithms such as word2vec and ELMo\nimprove the accuracy of many supervised NLP models, mainly because they can\ntake advantage of large amounts of unlabeled text. However, the supervised\nmodels only learn from task-specific labeled data during the main training\nphase. We therefore propose Cross-View Training (CVT), a semi-supervised\nlearning algorithm that improves the representations of a Bi-LSTM sentence\nencoder using a mix of labeled and unlabeled data. On labeled examples,\nstandard supervised learning is used. On unlabeled examples, CVT teaches\nauxiliary prediction modules that see restricted views of the input (e.g., only\npart of a sentence) to match the predictions of the full model seeing the whole\ninput. Since the auxiliary modules and the full model share intermediate\nrepresentations, this in turn improves the full model. Moreover, we show that\nCVT is particularly effective when combined with multi-task learning. We\nevaluate CVT on five sequence tagging tasks, machine translation, and\ndependency parsing, achieving state-of-the-art results.","url_abs":"http://arxiv.org/abs/1809.08370v1","url_pdf":"http://arxiv.org/pdf/1809.08370v1.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":"semi-supervised-sequence-modeling-with-cross","repo_url":"https://github.com/tensorflow/models","is_official":0,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":null},{"paper_slug":"semi-supervised-sequence-modeling-with-cross","repo_url":"https://github.com/rezkaaufar/pytorch-cvt","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"ccg-supertagging","task_name":"CCG Supertagging"},{"task_slug":"dependency-parsing","task_name":"Dependency Parsing"},{"task_slug":"machine-translation","task_name":"Machine Translation"},{"task_slug":"multi-task-learning","task_name":"Multi-Task Learning"},{"task_slug":"named-entity-recognition-ner","task_name":"Named Entity Recognition (NER)"},{"task_slug":"part-of-speech-tagging","task_name":"Part-Of-Speech Tagging"},{"task_slug":"representation-learning","task_name":"Representation Learning"},{"task_slug":"sentence","task_name":"Sentence"},{"task_slug":"translation","task_name":"Translation"}],"methods":[{"method_slug":"bilstm","method_name":"BiLSTM"},{"method_slug":"cnn-bilstm","method_name":"CNN BiLSTM"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"cross-view-training","method_name":"Cross-View Training"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"lstm","method_name":"LSTM"},{"method_slug":"sigmoid-activation","method_name":"Sigmoid Activation"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"tanh-activation","method_name":"Tanh Activation"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/ccg-supertagging-on-ccgbank","task":"CCG Supertagging","dataset":"CCGbank","model":"CVT + Multi-task + Large","rank_in_archive_order":3,"of":8,"metrics":{"Accuracy":"96.1"},"uses_additional_data":false},{"leaderboard":"/sota/dependency-parsing-on-penn-treebank","task":"Dependency Parsing","dataset":"Penn Treebank","model":"CVT + Multi-Task","rank_in_archive_order":7,"of":22,"metrics":{"LAS":"95.02","UAS":"96.61"},"uses_additional_data":false},{"leaderboard":"/sota/machine-translation-on-iwslt2015-english-1","task":"Machine Translation","dataset":"IWSLT2015 English-Vietnamese","model":"CVT","rank_in_archive_order":6,"of":11,"metrics":{"BLEU":"29.6"},"uses_additional_data":true},{"leaderboard":"/sota/named-entity-recognition-ner-on-conll-2003","task":"Named Entity Recognition (NER)","dataset":"CoNLL 2003 (English)","model":"CVT + Multi-Task","rank_in_archive_order":39,"of":73,"metrics":{"F1":"92.61"},"uses_additional_data":true},{"leaderboard":"/sota/named-entity-recognition-ner-on-conll-2003","task":"Named Entity Recognition (NER)","dataset":"CoNLL 2003 (English)","model":"CVT + Multi-Task + Large","rank_in_archive_order":40,"of":73,"metrics":{"F1":"92.61"},"uses_additional_data":false},{"leaderboard":"/sota/named-entity-recognition-ner-on-ontonotes-v5","task":"Named Entity Recognition (NER)","dataset":"Ontonotes v5 (English)","model":"CVT + Multi-Task + Large","rank_in_archive_order":18,"of":28,"metrics":{"F1":"88.81"},"uses_additional_data":false},{"leaderboard":"/sota/part-of-speech-tagging-on-penn-treebank","task":"Part-Of-Speech Tagging","dataset":"Penn Treebank","model":"CVT + Multi-task","rank_in_archive_order":5,"of":20,"metrics":{"Accuracy":"97.76"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1809.08370","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}