{"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/hg-means-a-scalable-hybrid-genetic-algorithm","title":"HG-means: A scalable hybrid genetic algorithm for minimum sum-of-squares clustering","arxiv_id":"1804.09813","date":"2018-04-25","proceeding":null,"authors":["Daniel Gribel","Thibaut Vidal"],"abstract":"Minimum sum-of-squares clustering (MSSC) is a widely used clustering model,\nof which the popular K-means algorithm constitutes a local minimizer. It is\nwell known that the solutions of K-means can be arbitrarily distant from the\ntrue MSSC global optimum, and dozens of alternative heuristics have been\nproposed for this problem. However, no other algorithm has been predominantly\nadopted in the literature. This may be related to differences of computational\neffort, or to the assumption that a near-optimal solution of the MSSC has only\na marginal impact on clustering validity. In this article, we dispute this\nbelief. We introduce an efficient population-based metaheuristic that uses\nK-means as a local search in combination with problem-tailored crossover,\nmutation, and diversification operators. This algorithm can be interpreted as a\nmulti-start K-means, in which the initial center positions are carefully\nsampled based on the search history. The approach is scalable and accurate,\noutperforming all recent state-of-the-art algorithms for MSSC in terms of\nsolution quality, measured by the depth of local minima. This enhanced accuracy\nleads to clusters which are significantly closer to the ground truth than those\nof other algorithms, for overlapping Gaussian-mixture datasets with a large\nnumber of features. Therefore, improved global optimization methods appear to\nbe essential to better exploit the MSSC model in high dimension.","url_abs":"http://arxiv.org/abs/1804.09813v2","url_pdf":"http://arxiv.org/pdf/1804.09813v2.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":"hg-means-a-scalable-hybrid-genetic-algorithm","repo_url":"https://github.com/danielgribel/hg-means","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"clustering","task_name":"Clustering"},{"task_slug":"global-optimization","task_name":"global-optimization"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}