From Supervised to Unsupervised Support Vector Machines and Applications in Astronomy

摘要: Support vector machines are among the most popular techniques in machine learning. Given sufficient labeled data, they often yield excellent results. However, for a variety of real-world tasks, acquisition data can be very time-consuming; unlabeled on other hand, obtained easily huge quantities. Semi-supervised support try to take advantage these additional patterns and have been successfully applied this context. induce hard combinatorial optimization problem. In work, we present two strategies that address task evaluate potential resulting implementations sets, including an example from field astronomy.