Transductive polyhedral conic classifiers Transdükdif çokyüzlü konik siniflandiricilar

Saglamlar H., ÇEVİKALP H.

29th IEEE Conference on Signal Processing and Communications Applications, SIU 2021, Virtual, Istanbul, Türkiye, 9 - 11 Haziran 2021 identifier identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Cilt numarası:
  • Doi Numarası: 10.1109/siu53274.2021.9477913
  • Basıldığı Şehir: Virtual, Istanbul
  • Basıldığı Ülke: Türkiye
  • Anahtar Kelimeler: polyhedral conic classifier, SVM, transductive support vector machine, semi-supervised learning
  • Eskişehir Osmangazi Üniversitesi Adresli: Evet

Özet

© 2021 IEEE.In this study, transductive (semi-supervised) polyhedral conic classifiers are proposed. In the proposed method, concave-convex procedure is used to solve the optimization problem as in Robust Transductive Support Vector Machines. This procedure decomposes a non-convex optimization problem into concave and convex parts. Unlike RTSVM that uses SVM formulation, our proposed method uses polyhedral conic classifiers (PCC) that provides tight and closed decision boundaries compared to SVM. In this way, effective, fast and better results can be obtained in large-scale datasets.