Supervised learning in presence of outliers, label noise and unobserved classes

Andrea Cappozzo, Francesca Greselin, Brendan Murphy

Risultato della ricerca: Contributo in libroContributo a convegno

Abstract

Three important issues are often encountered in Supervised Classifica- tion: class-memberships are unreliable for some training units (Label Noise), a pro- portion of observations might depart from the bulk of the data structure (Outliers) and groups represented in the test set may have not been encountered earlier in the learn- ing phase (Unobserved Classes). The present work introduces a Robust and Adaptive Eigenvalue-Decomposition Discriminant Analysis (RAEDDA) capable of handling situations in which one or more of the afore described problems occur. Transductive and inductive robust EM-based procedures are proposed for parameter estimation and experiments on real data, artificially adulterated, are provided to underline the benefits of the proposed method.
Lingua originaleEnglish
Titolo della pubblicazione ospiteCladag2019 : Book of short papers
Pagine104-107
Numero di pagine4
Stato di pubblicazionePubblicato - 2019
EventoScientific Meeting Classification and Data Analysis Group - Cassino
Durata: 11 set 201913 set 2019

Convegno

ConvegnoScientific Meeting Classification and Data Analysis Group
CittàCassino
Periodo11/9/1913/9/19

Keywords

  • model-based classification
  • unobserved classes
  • robust estimation
  • outliers detection
  • impartial trimming
  • label noise

Fingerprint

Entra nei temi di ricerca di 'Supervised learning in presence of outliers, label noise and unobserved classes'. Insieme formano una fingerprint unica.

Cita questo