Classification is a important field of study in Machine Learning area. Traditional algorithms relay on closed-set scenarios, where the label classes are known in prior. Thus, in real world we eventually deal with unknown instances of data. To treat unseen instances in a proper way, a approach called Openset classification was developed. This paper briefly surveys some of the recent works in Open-set Classification.
- Alan C. Neves
- Caio C. V. da Silva
- Jéssica Soares
- Sérgio J. de Sousa