Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning
This addresses the lack of uncertainty quantification in deep learning, which is crucial for applications like reinforcement learning, by providing a computationally efficient Bayesian approximation.
The paper tackles the problem of representing model uncertainty in deep neural networks by theoretically linking dropout training to approximate Bayesian inference in Gaussian processes, enabling uncertainty estimation without added computational cost or accuracy loss. It shows improvements in predictive log-likelihood and RMSE over state-of-the-art methods on tasks like regression and classification using MNIST.
Deep learning tools have gained tremendous attention in applied machine learning. However such tools for regression and classification do not capture model uncertainty. In comparison, Bayesian models offer a mathematically grounded framework to reason about model uncertainty, but usually come with a prohibitive computational cost. In this paper we develop a new theoretical framework casting dropout training in deep neural networks (NNs) as approximate Bayesian inference in deep Gaussian processes. A direct result of this theory gives us tools to model uncertainty with dropout NNs -- extracting information from existing models that has been thrown away so far. This mitigates the problem of representing uncertainty in deep learning without sacrificing either computational complexity or test accuracy. We perform an extensive study of the properties of dropout's uncertainty. Various network architectures and non-linearities are assessed on tasks of regression and classification, using MNIST as an example. We show a considerable improvement in predictive log-likelihood and RMSE compared to existing state-of-the-art methods, and finish by using dropout's uncertainty in deep reinforcement learning.