Subspace Inference for Bayesian Deep Learning

  1. Photo Pavel Izmailov Pavel Izmailov
  2. Photo Wesley J. Maddox Wesley J. Maddox
  3. Photo Polina Kirichenko Polina Kirichenko
  4. Photo Timur Garipov Timur Garipov
  5. Photo Dmitry Vetrov Dmitry Vetrov
  6. Photo Andrew Gordon Wilson Andrew Gordon Wilson

Bayesian inference was once a gold standard for learning with neural networks, providing accurate full predictive distributions and well calibrated uncertainty. However, scaling Bayesian inference techniques to deep neural networks is challenging due to the high dimensionality of the parameter space. In this paper, we construct low-dimensional subspaces of parameter space, such as the first principal components of the stochastic gradient descent (SGD) trajectory, which contain diverse sets of high performing models. In these subspaces, we are able to apply elliptical slice sampling and variational inference, which struggle in the full parameter space. We show that Bayesian model averaging over the induced posterior in these subspaces produces accurate predictions and well calibrated predictive uncertainty for both regression and image classification.