Further, we emphasize the significance of principled regularization of the network parameters and prediction.

However, practical posteriors are often, even locally, highly non-Gaussian, and empirical performance deteriorates.

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

First, it requires efficient and flexible parameterisations of layer-wise equivari-ances.

Due to the non-linearity of NNs, no analytic solution to the integral exists, even when the likelihood and the approximate posterior are both Gaussian. A low-cost, unbiased, stochastic approximation can be obtained via Monte Carlo (MC) integration: obtainS samples from the approximate posterior and then compute the empirical expectation of the likelihood w.r.t.