SVM Venn Machine with k-Means Clustering

摘要

In this paper, we introduce a new method of designing Venn Machine taxonomy based on Support Vector Machines and k-means clustering for both binary and multi-class problems. We compare this algorithm to some other multi-probabilistic predictors including SVM Venn Machine with homogeneous intervals and a recently developed algorithm called Venn-ABERS predictor. These algorithms were tested on a range of real-world data sets. Experimental results are presented and discussed.