{"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/variational-deep-semantic-hashing-for-text","title":"Variational Deep Semantic Hashing for Text Documents","arxiv_id":"1708.03436","date":"2017-08-11","proceeding":null,"authors":["Suthee Chaidaroon","Yi Fang"],"abstract":"As the amount of textual data has been rapidly increasing over the past\ndecade, efficient similarity search methods have become a crucial component of\nlarge-scale information retrieval systems. A popular strategy is to represent\noriginal data samples by compact binary codes through hashing. A spectrum of\nmachine learning methods have been utilized, but they often lack expressiveness\nand flexibility in modeling to learn effective representations. The recent\nadvances of deep learning in a wide range of applications has demonstrated its\ncapability to learn robust and powerful feature representations for complex\ndata. Especially, deep generative models naturally combine the expressiveness\nof probabilistic generative models with the high capacity of deep neural\nnetworks, which is very suitable for text modeling. However, little work has\nleveraged the recent progress in deep learning for text hashing.\n  In this paper, we propose a series of novel deep document generative models\nfor text hashing. The first proposed model is unsupervised while the second one\nis supervised by utilizing document labels/tags for hashing. The third model\nfurther considers document-specific factors that affect the generation of\nwords. The probabilistic generative formulation of the proposed models provides\na principled framework for model extension, uncertainty estimation, simulation,\nand interpretability. Based on variational inference and reparameterization,\nthe proposed models can be interpreted as encoder-decoder deep neural networks\nand thus they are capable of learning complex nonlinear distributed\nrepresentations of the original documents. We conduct a comprehensive set of\nexperiments on four public testbeds. The experimental results have demonstrated\nthe effectiveness of the proposed supervised learning models for text hashing.","url_abs":"http://arxiv.org/abs/1708.03436v1","url_pdf":"http://arxiv.org/pdf/1708.03436v1.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":"variational-deep-semantic-hashing-for-text","repo_url":"https://github.com/J-zin/SNUH","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"variational-deep-semantic-hashing-for-text","repo_url":"https://github.com/MindSpore-scientific/code-9/tree/main/SNUH","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null},{"paper_slug":"variational-deep-semantic-hashing-for-text","repo_url":"https://github.com/unsuthee/VariationalDeepSemanticHashing","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"tf","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"image-retrieval","task_name":"Image Retrieval"},{"task_slug":"information-retrieval","task_name":"Information Retrieval"},{"task_slug":"retrieval","task_name":"Retrieval"},{"task_slug":"supervised-text-retrieval","task_name":"Supervised Text Retrieval"},{"task_slug":"text-retrieval","task_name":"Text Retrieval"},{"task_slug":"variational-inference","task_name":"Variational Inference"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/text-retrieval-on-20-newsgroups","task":"Text Retrieval","dataset":"20 Newsgroups","model":"VDSH","rank_in_archive_order":2,"of":3,"metrics":{"Precision@100":"0.4354"},"uses_additional_data":false},{"leaderboard":"/sota/text-retrieval-on-reuters-21578","task":"Text Retrieval","dataset":"Reuters-21578","model":"VDSH","rank_in_archive_order":1,"of":3,"metrics":{"Precision@100":"0.7753"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1708.03436","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}