Asymmetric Transfer Learning with Deep Gaussian Processes

We introduce a novel Gaussian process based Bayesian model for asymmetric transfer learn-ing. We adopt a two-layer feed-forward deep Gaussian process as the task learner of source and target domains. The first layer projects the data onto a separate non-linear manifold for each task. We perform knowledge transfer by pro-jecting the target data also onto the source do-main and linearly combining its representations on the source and target domain manifolds. Our approach achieves the state-of-the-art in a bench-mark real-world image categorization task, and improves on it in cross-tissue tumor detection from histopathology tissue slide images