SVM solves multi-class problems using pairwise classification (or 1-vs-1). This means that the predicted probabilities are coupled using Hastie and Tibshirani's pairwise coupling method.
Pairwise coupling is a popular multi-class classification method that combines all comparisons for each pair of classes. This paper presents two approaches for obtaining class probabilities. Both methods can be reduced to linear systems and are easy to implement. It shows conceptually and experimentally that the proposed approaches are more stable than the two existing popular methods: voting and the method by Hastie and Tibshirani.