{"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/getting-deep-recommenders-fit-bloom","title":"Getting deep recommenders fit: Bloom embeddings for sparse binary input/output networks","arxiv_id":"1706.03993","date":"2017-06-13","proceeding":null,"authors":["Joan Serrà","Alexandros Karatzoglou"],"abstract":"Recommendation algorithms that incorporate techniques from deep learning are\nbecoming increasingly popular. Due to the structure of the data coming from\nrecommendation domains (i.e., one-hot-encoded vectors of item preferences),\nthese algorithms tend to have large input and output dimensionalities that\ndominate their overall size. This makes them difficult to train, due to the\nlimited memory of graphical processing units, and difficult to deploy on mobile\ndevices with limited hardware. To address these difficulties, we propose Bloom\nembeddings, a compression technique that can be applied to the input and output\nof neural network models dealing with sparse high-dimensional binary-coded\ninstances. Bloom embeddings are computationally efficient, and do not seriously\ncompromise the accuracy of the model up to 1/5 compression ratios. In some\ncases, they even improve over the original accuracy, with relative increases up\nto 12%. We evaluate Bloom embeddings on 7 data sets and compare it against 4\nalternative methods, obtaining favorable results. We also discuss a number of\nfurther advantages of Bloom embeddings, such as 'on-the-fly' constant-time\noperation, zero or marginal space requirements, training time speedups, or the\nfact that they do not require any change to the core model architecture or\ntraining configuration.","url_abs":"http://arxiv.org/abs/1706.03993v1","url_pdf":"http://arxiv.org/pdf/1706.03993v1.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":"getting-deep-recommenders-fit-bloom","repo_url":"https://github.com/expediagroup/kamae","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1706.03993","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}