While possessing high representational capacity, these architectures primarily operate on grid-like or sequential structures due to their built-in "innate priors".

Due to the non-convex nature of such joint estimators, the theoretical justification of their efficiency is typically challenging.

As a result, the brain needs to perform these computations in an approximate manner.

Wenwu Zhu is the corresponding author.

Neural Information Processing Systems http://nips.cc/

Many real-world vision problems suffer from inherent ambiguities. In clinical applications for example, it might not be clear from a CT scan alone which particular region is cancer tissue. Therefore a group of graders typically produces a set of diverse but plausible segmentations.

Neural Information Processing Systems http://nips.cc/

We propose DropMax, a stochastic version of softmax classifier which at each iteration drops non-target classes according to dropout probabilities adaptively decided for each instance.

Deep reinforcement learning methods traditionally struggle with tasks where environment rewards are particularly sparse.