A comprehensive revisit to one class classification
Qi Wang, Jincai Huang, Yanghe Feng, Zhihao Luo, Changjun Fan, Xingxing Liang · 2017
The one class classification (OCC), a special learning framework dealing with the positive and unlabeled dataset, has ubiquitous applications, especially in the era of information-explosion. Since it has aroused researchers' interests from both academia and industry, we provide a comprehensive revisit to algorithms for the OCC from the perspective of methodologies with the ideas of essence behind common-used algorithms uncovered and analyzed in depth. Besides, it is the first time to illustrate the challenges in the domain while some novel ideas are suggested here for further trials or development. Some employments are also involved in the paper and characteristics of methodologies are talked about for broadening the scope of applications. Finally, several topics worthy to research are mentioned in the guideline of literature study.