Reliable training and estimation of variance networks

Abstract

We propose and investigate new complementary methodologies for estimating predictive variance networks in regression neural networks. We derive a locally aware mini-batching scheme that result in sparse robust gradients, and show how to make unbiased weight updates to a variance network. Further, we formulate a heuristic for robustly fitting both the mean and variance networks post hoc. Finally, we take inspiration from posterior Gaussian processes and propose a network architecture with similar extrapolation properties to Gaussian processes. The proposed methodologies are complementary, and improve upon baseline methods individually. Experimentally, we investigate the impact on predictive uncertainty on multiple datasets and tasks ranging from regression, active learning and generative modeling. Experiments consistently show significant improvements in predictive uncertainty estimation over state-of-the-art methods across tasks and datasets.

arXiv commentAppeared at NeurIPS 2019

References in corpus (11)

• Auto-Encoding Variational Bayes
• Improving neural networks by preventing co-adaptation of feature detectors
• What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?
• On Calibration of Modern Neural Networks
• Probabilistic Backpropagation for Scalable Learning of Bayesian Neural Networks
• A Tutorial on Bayesian Optimization
• Flipout: Efficient Pseudo-Independent Weight Perturbations on Mini-Batches
• A Simple Baseline for Bayesian Uncertainty in Deep Learning
• High-Quality Prediction Intervals for Deep Learning: A Distribution-Free, Ensembled Approach
• EFANNA : An Extremely Fast Approximate Nearest Neighbor Search Algorithm Based on kNN Graph
• Only Bayes should learn a manifold (on the estimation of differential geometric structure from data)

Cited by in corpus (8)

• Uncertainty Toolbox: an Open-Source Library for Assessing, Visualizing, and Improving Uncertainty Quantification
• Only Bayes should learn a manifold (on the estimation of differential geometric structure from data)
• Beyond Pinball Loss: Quantile Methods for Calibrated Uncertainty Quantification
• Variational Autoencoders with Riemannian Brownian Motion Priors
• Estimating and Evaluating Regression Predictive Uncertainty in Deep Object Detectors
• Variational Variance: Simple, Reliable, Calibrated Heteroscedastic Noise Variance Parameterization
• Exploring Uncertainty in Deep Learning for Construction of Prediction Intervals
• Addressing Variance Shrinkage in Variational Autoencoders using Quantile Regression