Van der Walt, Christiaan MBarnard, E2017-09-182017-09-182017-02Van der Walt, C.M. & Barnard, E. 2017. Variable kernel density estimation in high-dimensional feature spaces. Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence (AAAI), 4-9 February 2017, San Francisco, California, USAhttps://aaai.org/ocs/index.php/AAAI/AAAI17/paper/view/14737http://hdl.handle.net/10204/9562Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence (AAAI), 4-9 February 2017, San Francisco, California, USAEstimating the joint probability density function of a dataset is a central task in many machine learning applications. In this work we address the fundamental problem of kernel bandwidth estimation for variable kernel density estimation in high-dimensional feature spaces. We derive a variable kernel bandwidth estimator by minimizing the leave-one-out entropy objective function and show that this estimator is capable of performing estimation in high-dimensional feature spaces with great success. We compare the performance of this estimator to state-of-the art maximum-likelihood estimators on a number of representative high-dimensional machine learning tasks and show that the newly introduced minimum leave-one-out entropy estimator performs optimally on a number of high-dimensional datasets considered.enMachine learningProbability density estimationNon-parametric density estimationKernel bandwidth estimationMaximum-likelihoodConference PresentationVan der Walt, C. M., & Barnard, E. (2017). Variable kernel density estimation in high-dimensional feature spaces. Association for the Advancement of Artificial. http://hdl.handle.net/10204/9562Van der Walt, Christiaan M, and E Barnard. "Variable kernel density estimation in high-dimensional feature spaces." (2017): http://hdl.handle.net/10204/9562Van der Walt CM, Barnard E, Variable kernel density estimation in high-dimensional feature spaces; Association for the Advancement of Artificial; 2017. http://hdl.handle.net/10204/9562 .TY - Conference Presentation AU - Van der Walt, Christiaan M AU - Barnard, E AB - Estimating the joint probability density function of a dataset is a central task in many machine learning applications. In this work we address the fundamental problem of kernel bandwidth estimation for variable kernel density estimation in high-dimensional feature spaces. We derive a variable kernel bandwidth estimator by minimizing the leave-one-out entropy objective function and show that this estimator is capable of performing estimation in high-dimensional feature spaces with great success. We compare the performance of this estimator to state-of-the art maximum-likelihood estimators on a number of representative high-dimensional machine learning tasks and show that the newly introduced minimum leave-one-out entropy estimator performs optimally on a number of high-dimensional datasets considered. DA - 2017-02 DB - ResearchSpace DP - CSIR KW - Machine learning KW - Probability density estimation KW - Non-parametric density estimation KW - Kernel bandwidth estimation KW - Maximum-likelihood LK - https://researchspace.csir.co.za PY - 2017 T1 - Variable kernel density estimation in high-dimensional feature spaces TI - Variable kernel density estimation in high-dimensional feature spaces UR - http://hdl.handle.net/10204/9562 ER -