ujohn33 / rvm_project Goto Github PK
View Code? Open in Web Editor NEWAdvnaced Machine learning research project, which I contributed to. In this project we explore the general Bayesian framework for obtaining sparse solutions first suggested by Michael E. Tipping. A special case application in this framework known as relevance vector machine (RVM) was implemented. The method is compared to a well-known support vector machine technique. The results are compared for classification and regression problems for artificially generated and real data sets and compared to equivalent results reported in the original paper by Tipping. Though our results generally aligned with Tipping's, we offer possible causes for the discrepancies in the results, whether they are regarding precision, or in the general features of RVM.