Classification with Lorentzian Distance Metric

BİLGE H. Ş., Kerimbekov Y.

23nd Signal Processing and Communications Applications Conference (SIU), Malatya, Türkiye, 16 - 19 Mayıs 2015, ss.2106-2109 identifier identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Cilt numarası:
  • Doi Numarası: 10.1109/siu.2015.7130286
  • Basıldığı Şehir: Malatya
  • Basıldığı Ülke: Türkiye
  • Sayfa Sayıları: ss.2106-2109
  • Gazi Üniversitesi Adresli: Evet

Özet

In this study, a new distance metric that can be used in classification is proposed and its success is investigated. The classification success is increased by using the special distance metric in Lorentzian space. In order to find the optimum angle in transformation from Euclidean space to Lorentzian space, the decision line from Support Vector Machine is utilized. In experimental studies, the success of the proposed method is investigated by using open source data sets; WINE, ECOLI, BLOGGER, DIABET.