{"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/deep-learning-over-multi-field-categorical","title":"Deep Learning over Multi-field Categorical Data: A Case Study on User Response Prediction","arxiv_id":"1601.02376","date":"2016-01-11","proceeding":null,"authors":["Wei-Nan Zhang","Tianming Du","Jun Wang"],"abstract":"Predicting user responses, such as click-through rate and conversion rate,\nare critical in many web applications including web search, personalised\nrecommendation, and online advertising. Different from continuous raw features\nthat we usually found in the image and audio domains, the input features in web\nspace are always of multi-field and are mostly discrete and categorical while\ntheir dependencies are little known. Major user response prediction models have\nto either limit themselves to linear models or require manually building up\nhigh-order combination features. The former loses the ability of exploring\nfeature interactions, while the latter results in a heavy computation in the\nlarge feature space. To tackle the issue, we propose two novel models using\ndeep neural networks (DNNs) to automatically learn effective patterns from\ncategorical feature interactions and make predictions of users' ad clicks. To\nget our DNNs efficiently work, we propose to leverage three feature\ntransformation methods, i.e., factorisation machines (FMs), restricted\nBoltzmann machines (RBMs) and denoising auto-encoders (DAEs). This paper\npresents the structure of our models and their efficient training algorithms.\nThe large-scale experiments with real-world data demonstrate that our methods\nwork better than major state-of-the-art models.","url_abs":"http://arxiv.org/abs/1601.02376v1","url_pdf":"http://arxiv.org/pdf/1601.02376v1.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":"deep-learning-over-multi-field-categorical","repo_url":"https://github.com/wnzhang/deep-ctr","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"deep-learning-over-multi-field-categorical","repo_url":"https://github.com/UlionTse/mlgb","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"deep-learning-over-multi-field-categorical","repo_url":"https://github.com/ddatta-DAC/Learning","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"deep-learning-over-multi-field-categorical","repo_url":"https://github.com/shenweichen/DeepCTR","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"deep-learning-over-multi-field-categorical","repo_url":"https://github.com/shenweichen/DeepCTR-Torch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"click-through-rate-prediction","task_name":"Click-Through Rate Prediction"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/click-through-rate-prediction-on-company","task":"Click-Through Rate Prediction","dataset":"Company*","model":"FNN","rank_in_archive_order":2,"of":8,"metrics":{"AUC":"0.8683","Log Loss":"0.02629"},"uses_additional_data":false},{"leaderboard":"/sota/click-through-rate-prediction-on-criteo","task":"Click-Through Rate Prediction","dataset":"Criteo","model":"FNN","rank_in_archive_order":39,"of":39,"metrics":{"AUC":"0.7963","Log Loss":"0.45738"},"uses_additional_data":false},{"leaderboard":"/sota/click-through-rate-prediction-on-ipinyou","task":"Click-Through Rate Prediction","dataset":"iPinYou","model":"FNN","rank_in_archive_order":7,"of":7,"metrics":{"AUC":"0.7619"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1601.02376","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}