Аннотация
We introduce ‘semi-unsupervised learning’, a problem regime related to transferlearning and zero/few shot learning where, in the training data, some classes aresparsely labelled and others entirely unlabelled. Models able to learn from trainingdata of this type are potentially of great use as many real-world datasets are likethis. Here we demonstrate a new deep generative model for classification in thisregime. Our model, a Gaussian mixture deep generative model, demonstratessuperior semi-unsupervised classification performance on MNIST to model M2from Kingma and Welling (2014).
Пользователи данного ресурса
Пожалуйста,
войдите в систему, чтобы принять участие в дискуссии (добавить собственные рецензию, или комментарий)