Goto

Collaborating Authors

 well-written paper


Reviews: Stochastic Variational Deep Kernel Learning

Neural Information Processing Systems

This is a well-written paper with only a small number of grammatical and presentation errors, e.g. However, the experimental section lacks some detail on the training of DNNs and the pre-training. From the level of detail given it will be hard to reproduce the results. Furthermore the results are not too convincing. The authors claim that "flexible deep kernel GPs achieve superior performance" to DNNs and SVMs.