{"id":819,"date":"2018-07-21T06:30:43","date_gmt":"2018-07-21T06:30:43","guid":{"rendered":"https:\/\/www.aiproblog.com\/index.php\/2018\/07\/21\/feature-selection-for-unsupervised-learning\/"},"modified":"2018-07-21T06:30:43","modified_gmt":"2018-07-21T06:30:43","slug":"feature-selection-for-unsupervised-learning","status":"publish","type":"post","link":"https:\/\/www.aiproblog.com\/index.php\/2018\/07\/21\/feature-selection-for-unsupervised-learning\/","title":{"rendered":"Feature Selection For Unsupervised Learning"},"content":{"rendered":"

Author: Vincent Granville<\/p>\n

\n

This is my presentation for the IBM data science day<\/a>, July 24.<\/em><\/p>\n

Abstract<\/strong><\/p>\n

After reviewing popular techniques used in supervised, unsupervised and semi-supervised machine learning, we focus on feature selection methods in these different contexts, especially the metrics used to assess the value of a feature or set of features, be it binary, continuous or categorical variables. <\/p>\n

We go in deeper details and review modern feature selection techniques for unsupervised learning, typically relying on entropy-like criteria. While these criteria are usually model-dependent or scale-dependent, we introduce a new model-free, data-driven methodology in this context, with an application to an interesting number theory problem (simulated data set) in which each feature has a known theoretical entropy. <\/p>\n

We also briefly discuss high precision computing as it is relevant to this peculiar data set, as well as units of information smaller than the bit.<\/p>\n

To download the presentation, click here<\/a>\u00a0(PowerPoint document.)<\/p>\n

DSC Resources<\/b><\/span><\/p>\n