{"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/conditional-random-field-autoencoders-for","title":"Conditional Random Field Autoencoders for Unsupervised Structured Prediction","arxiv_id":"1411.1147","date":"2014-11-05","proceeding":"NeurIPS 2014 12","authors":["Waleed Ammar","Chris Dyer","Noah A. Smith"],"abstract":"We introduce a framework for unsupervised learning of structured predictors\nwith overlapping, global features. Each input's latent representation is\npredicted conditional on the observable data using a feature-rich conditional\nrandom field. Then a reconstruction of the input is (re)generated, conditional\non the latent structure, using models for which maximum likelihood estimation\nhas a closed-form. Our autoencoder formulation enables efficient learning\nwithout making unrealistic independence assumptions or restricting the kinds of\nfeatures that can be used. We illustrate insightful connections to traditional\nautoencoders, posterior regularization and multi-view learning. We show\ncompetitive results with instantiations of the model for two canonical NLP\ntasks: part-of-speech induction and bitext word alignment, and show that\ntraining our model can be substantially more efficient than comparable\nfeature-rich baselines.","url_abs":"http://arxiv.org/abs/1411.1147v2","url_pdf":"http://arxiv.org/pdf/1411.1147v2.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":"conditional-random-field-autoencoders-for","repo_url":"https://github.com/ldmt-muri/alignment-with-openfst","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"multi-view-learning","task_name":"MULTI-VIEW LEARNING"},{"task_slug":"prediction","task_name":"Prediction"},{"task_slug":"structured-prediction","task_name":"Structured Prediction"},{"task_slug":"word-alignment","task_name":"Word Alignment"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1411.1147","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}