{"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/afs-an-attention-based-mechanism-for","title":"AFS: An Attention-based mechanism for Supervised Feature Selection","arxiv_id":"1902.11074","date":"2019-02-28","proceeding":null,"authors":["Ning Gui","Danni Ge","Ziyin Hu"],"abstract":"As an effective data preprocessing step, feature selection has shown its\neffectiveness to prepare high-dimensional data for many machine learning tasks.\nThe proliferation of high di-mension and huge volume big data, however, has\nbrought major challenges, e.g. computation complexity and stability on noisy\ndata, upon existing feature-selection techniques. This paper introduces a novel\nneural network-based feature selection architecture, dubbed Attention-based\nFeature Selec-tion (AFS). AFS consists of two detachable modules: an at-tention\nmodule for feature weight generation and a learning module for the problem\nmodeling. The attention module for-mulates correlation problem among features\nand supervision target into a binary classification problem, supported by a\nshallow attention net for each feature. Feature weights are generated based on\nthe distribution of respective feature se-lection patterns adjusted by\nbackpropagation during the train-ing process. The detachable structure allows\nexisting off-the-shelf models to be directly reused, which allows for much less\ntraining time, demands for the training data and requirements for expertise. A\nhybrid initialization method is also intro-duced to boost the selection\naccuracy for datasets without enough samples for feature weight generation.\nExperimental results show that AFS achieves the best accuracy and stability in\ncomparison to several state-of-art feature selection algo-rithms upon both\nMNIST, noisy MNIST and several datasets with small samples.","url_abs":"http://arxiv.org/abs/1902.11074v1","url_pdf":"http://arxiv.org/pdf/1902.11074v1.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":"afs-an-attention-based-mechanism-for","repo_url":"https://github.com/upup123/AAAI-2019-AFS","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok"}}],"tasks":[{"task_slug":"binary-classification","task_name":"Binary Classification"},{"task_slug":"feature-selection","task_name":"feature selection"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1902.11074","atlas_url":"https://app.syntology.ai/?focus=1902.11074","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}