Error correction output codes: Overview, challenges and future trends Hata Düzeltme Çikti Kodlari: Genel Bakiş, Zorluklar ve Gelecek Yönelimler
27th Signal Processing and Communications Applications Conference, SIU 2019, Sivas, Türkiye, 24 - 26 Nisan 2019, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/siu.2019.8806537
- Basıldığı Şehir: Sivas
- Basıldığı Ülke: Türkiye
- Anahtar Kelimeler: Error Correction Coding, Machine learning, Multi-class classification
- Boğaziçi Üniversitesi Adresli: Hayır
Özet
One of the most effective way to address multiclass classification problem is to use a set of judiciously designed binary classifiers and to carefully combine their results. Error-Correcting Output Codes (ECOC) is one of the successful frameworks that allows a division of labour through multiple binary classifications. This paper provides a brief introduction to the state-of-theart ECOC types, various decoding methods that merges binary classification results, and a comparative study that lays out challenges, advantages and disadvantages. We also provide few important applications of ECOC, its performance on MNIST data set and some of the future trends.