Generalisations of stochastic supervision models

Abstract When the labelling information is not deterministic, traditional supervised learning algorithms cannot be applied. In this case, stochastic supervision models provide a valuable alternative to classification. However, these are restricted in several aspects, which critically limits their applicability. paper, we four generalisations of models, extending them asymmetric assessments, multiple classes, feature-dependent assessments and multi-modal respectively. Corresponding generalisations, derive new EM algorithms. We show effectiveness our through illustrative examples simulated datasets, as well real-world three famous MNIST dataset, CIFAR-10 dataset EMNIST dataset.