Ensembles for Uncertainty Estimation: Benefits of Prior Functions and Bootstrapping
This work addresses uncertainty estimation for agents in ML, but it is incremental as it clarifies conflicting views on existing ensemble ingredients.
The paper tackled the problem of uncertainty estimation in machine learning by evaluating the benefits of prior functions and bootstrapping in ensemble methods, showing that prior functions improve joint predictions and bootstrapping helps when signal-to-noise ratios vary, with theoretical and experimental support.
In machine learning, an agent needs to estimate uncertainty to efficiently explore and adapt and to make effective decisions. A common approach to uncertainty estimation maintains an ensemble of models. In recent years, several approaches have been proposed for training ensembles, and conflicting views prevail with regards to the importance of various ingredients of these approaches. In this paper, we aim to address the benefits of two ingredients -- prior functions and bootstrapping -- which have come into question. We show that prior functions can significantly improve an ensemble agent's joint predictions across inputs and that bootstrapping affords additional benefits if the signal-to-noise ratio varies across inputs. Our claims are justified by both theoretical and experimental results.