A Simple Baseline for Bayesian Uncertainty in Deep Learning
This addresses uncertainty estimation for deep learning practitioners, offering a scalable and general-purpose solution that is incremental, building on existing Stochastic Weight Averaging.
The paper tackles the problem of uncertainty representation and calibration in deep learning by proposing SWA-Gaussian (SWAG), a method that approximates a Bayesian posterior over neural network weights using SGD iterates, resulting in improved performance on tasks like out-of-sample detection and calibration compared to alternatives.
We propose SWA-Gaussian (SWAG), a simple, scalable, and general purpose approach for uncertainty representation and calibration in deep learning. Stochastic Weight Averaging (SWA), which computes the first moment of stochastic gradient descent (SGD) iterates with a modified learning rate schedule, has recently been shown to improve generalization in deep learning. With SWAG, we fit a Gaussian using the SWA solution as the first moment and a low rank plus diagonal covariance also derived from the SGD iterates, forming an approximate posterior distribution over neural network weights; we then sample from this Gaussian distribution to perform Bayesian model averaging. We empirically find that SWAG approximates the shape of the true posterior, in accordance with results describing the stationary distribution of SGD iterates. Moreover, we demonstrate that SWAG performs well on a wide variety of tasks, including out of sample detection, calibration, and transfer learning, in comparison to many popular alternatives including MC dropout, KFAC Laplace, SGLD, and temperature scaling.