{"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/learning-latent-representations-in-neural","title":"Learning Latent Representations in Neural Networks for Clustering through Pseudo Supervision and Graph-based Activity Regularization","arxiv_id":"1802.03063","date":"2018-02-08","proceeding":"ICLR 2018 1","authors":["Ozsel Kilinc","Ismail Uysal"],"abstract":"In this paper, we propose a novel unsupervised clustering approach exploiting\nthe hidden information that is indirectly introduced through a pseudo\nclassification objective. Specifically, we randomly assign a pseudo\nparent-class label to each observation which is then modified by applying the\ndomain specific transformation associated with the assigned label. Generated\npseudo observation-label pairs are subsequently used to train a neural network\nwith Auto-clustering Output Layer (ACOL) that introduces multiple softmax nodes\nfor each pseudo parent-class. Due to the unsupervised objective based on\nGraph-based Activity Regularization (GAR) terms, softmax duplicates of each\nparent-class are specialized as the hidden information captured through the\nhelp of domain specific transformations is propagated during training.\nUltimately we obtain a k-means friendly latent representation. Furthermore, we\ndemonstrate how the chosen transformation type impacts performance and helps\npropagate the latent information that is useful in revealing unknown clusters.\nOur results show state-of-the-art performance for unsupervised clustering tasks\non MNIST, SVHN and USPS datasets, with the highest accuracies reported to date\nin the literature.","url_abs":"http://arxiv.org/abs/1802.03063v1","url_pdf":"http://arxiv.org/pdf/1802.03063v1.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":[],"tasks":[{"task_slug":"clustering","task_name":"Clustering"},{"task_slug":"unsupervised-image-classification","task_name":"Unsupervised Image Classification"}],"methods":[{"method_slug":"softmax","method_name":"Softmax"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/unsupervised-image-classification-on-mnist","task":"Unsupervised Image Classification","dataset":"MNIST","model":"ACOL + GAR + k-means","rank_in_archive_order":2,"of":10,"metrics":{"Accuracy":"98.32"},"uses_additional_data":false},{"leaderboard":"/sota/unsupervised-image-classification-on-svhn","task":"Unsupervised Image Classification","dataset":"SVHN","model":"ACOL-GAR","rank_in_archive_order":1,"of":4,"metrics":{"# of clusters (k)":"10","Acc":"76.80"},"uses_additional_data":true}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1802.03063","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}