Marginal Uncertainty in Linear Models and Deep Neural Networks - 42Papers