Dropout Strikes Back: Improved Uncertainty Estimation via Diversity Sampling

TL;DR

This paper introduces a novel sampling method for dropout layers that enhances uncertainty estimation in neural networks by selecting diverse neurons, leading to improved confidence and out-of-distribution detection without altering model training.

Contribution

It proposes a data-driven, diversity-based sampling approach for dropout that improves uncertainty estimation without modifying existing models or training procedures.

Findings

Achieves state-of-the-art uncertainty estimation results.

Effective for both regression and classification tasks.

Does not require changes to model architecture or training.

Abstract

Uncertainty estimation for machine learning models is of high importance in many scenarios such as constructing the confidence intervals for model predictions and detection of out-of-distribution or adversarially generated points. In this work, we show that modifying the sampling distributions for dropout layers in neural networks improves the quality of uncertainty estimation. Our main idea consists of two main steps: computing data-driven correlations between neurons and generating samples, which include maximally diverse neurons. In a series of experiments on simulated and real-world data, we demonstrate that the diversification via determinantal point processes-based sampling achieves state-of-the-art results in uncertainty estimation for regression and classification tasks. An important feature of our approach is that it does not require any modification to the models or training procedures, allowing straightforward application to any deep learning model with dropout layers.