{"title":"A method for searching for a globally optimal k-partition of higher-dimensional datasets","authors":"","doi":"10.1007/s10898-024-01372-6","DOIUrl":null,"url":null,"abstract":"<h3>Abstract</h3> <p>The problem of finding a globally optimal <em>k</em>-partition of a set <span> <span>\\(\\mathcal {A}\\)</span> </span> is a very intricate optimization problem for which in general, except in the case of one-dimensional data, i.e., for data with one feature (<span> <span>\\(\\mathcal {A}\\subset \\mathbb {R}\\)</span> </span>), there is no method to solve. Only in the one-dimensional case, there are efficient methods based on the fact that the search for a globally optimal <em>k</em>-partition is equivalent to solving a global optimization problem for a symmetric Lipschitz-continuous function using the global optimization algorithm <span>DIRECT</span>. In the present paper, we propose a method for finding a globally optimal <em>k</em>-partition in the general case (<span> <span>\\(\\mathcal {A}\\subset \\mathbb {R}^n\\)</span> </span>, <span> <span>\\(n\\ge 1\\)</span> </span>), generalizing an idea for solving the Lipschitz global optimization for symmetric functions. To do this, we propose a method that combines a global optimization algorithm with linear constraints and the <em>k</em>-means algorithm. The first of these two algorithms is used only to find a good initial approximation for the <em>k</em>-means algorithm. The method was tested on a number of artificial datasets and on several examples from the UCI Machine Learning Repository, and an application in spectral clustering for linearly non-separable datasets is also demonstrated. Our proposed method proved to be very efficient. </p>","PeriodicalId":15961,"journal":{"name":"Journal of Global Optimization","volume":"25 1","pages":""},"PeriodicalIF":1.8000,"publicationDate":"2024-02-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Global Optimization","FirstCategoryId":"100","ListUrlMain":"https://doi.org/10.1007/s10898-024-01372-6","RegionNum":3,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"Mathematics","Score":null,"Total":0}
引用次数: 0
Abstract
The problem of finding a globally optimal k-partition of a set \(\mathcal {A}\) is a very intricate optimization problem for which in general, except in the case of one-dimensional data, i.e., for data with one feature (\(\mathcal {A}\subset \mathbb {R}\)), there is no method to solve. Only in the one-dimensional case, there are efficient methods based on the fact that the search for a globally optimal k-partition is equivalent to solving a global optimization problem for a symmetric Lipschitz-continuous function using the global optimization algorithm DIRECT. In the present paper, we propose a method for finding a globally optimal k-partition in the general case (\(\mathcal {A}\subset \mathbb {R}^n\), \(n\ge 1\)), generalizing an idea for solving the Lipschitz global optimization for symmetric functions. To do this, we propose a method that combines a global optimization algorithm with linear constraints and the k-means algorithm. The first of these two algorithms is used only to find a good initial approximation for the k-means algorithm. The method was tested on a number of artificial datasets and on several examples from the UCI Machine Learning Repository, and an application in spectral clustering for linearly non-separable datasets is also demonstrated. Our proposed method proved to be very efficient.
期刊介绍:
The Journal of Global Optimization publishes carefully refereed papers that encompass theoretical, computational, and applied aspects of global optimization. While the focus is on original research contributions dealing with the search for global optima of non-convex, multi-extremal problems, the journal’s scope covers optimization in the widest sense, including nonlinear, mixed integer, combinatorial, stochastic, robust, multi-objective optimization, computational geometry, and equilibrium problems. Relevant works on data-driven methods and optimization-based data mining are of special interest.
In addition to papers covering theory and algorithms of global optimization, the journal publishes significant papers on numerical experiments, new testbeds, and applications in engineering, management, and the sciences. Applications of particular interest include healthcare, computational biochemistry, energy systems, telecommunications, and finance. Apart from full-length articles, the journal features short communications on both open and solved global optimization problems. It also offers reviews of relevant books and publishes special issues.