Mathematical Optimization approaches to supervised classification with noise labels

Il  Prof. Justo Puerto (IMUS, Universidad de Sevilla) terrà il seminario dal titolo: "Mathematical Optimization approaches to supervised classification with noise labels".   in modalità mista (online e on-site) in data giovedì 15 settembre 2022 ore 12:00 (stanza 24, 4° piano, Dipartimento di Scienze Statistiche).   Abstract:  In this talk we propose novel methodologies to optimally construct Support Vector Machines and classification trees based classifiers that take into account that label noise occurs in the training sample. We propose different alternatives based on solving Mixed Integer Linear and Non Linear models by incorporating decisions on relabeling some of the observations in the training dataset. The resulting method incorporates relabeling directly in the SVM or classification tree models while a second family of methods combines clustering with  classification at the same time, giving rise to models that apply simultaneously similarity measures and SVM or tree classifiers. Extensive computational experiments are reported based on a battery of standard datasets taken from UCI Machine Learning repository, showing the effectiveness of the proposed approaches.   Link all'evento: https://uniroma1.zoom.us/j/81728964719?pwd=aU1sUGpnSW5oRHdFYWVGUzRXSkErdz09 Il seminario è aperto a tutti gli interessati.
Relatore: 
Prof. Justo Puerto (IMUS, Universidad d
Data: 
15/09/2022 - 12:00
Luogo: 
[Modalità mista ]