Interpretable Uncertainty Quantification in AI for HEP

Abstract

Estimating uncertainty is at the core of performing scientific measurements in HEP: a measurement is not useful without an estimate of its uncertainty. The goal of uncertainty quantification (UQ) is inextricably linked to the question, "how do we physically and statistically interpret these uncertainties?" The answer to this question depends not only on the computational task we aim to undertake, but also on the methods we use for that task. For artificial intelligence (AI) applications in HEP, there are several areas where interpretable methods for UQ are essential, including inference, simulation, and control/decision-making. There exist some methods for each of these areas, but they have not yet been demonstrated to be as trustworthy as more traditional approaches currently employed in physics (e.g., non-AI frequentist and Bayesian methods). Shedding light on the questions above requires additional understanding of the interplay of AI systems and uncertainty quantification. We briefly discuss the existing methods in each area and relate them to tasks across HEP. We then discuss recommendations for avenues to pursue to develop the necessary techniques for reliable widespread usage of AI with UQ over the next decade.

arXiv commentSubmitted to the Proceedings of the US Community Study on the Future of Particle Physics (Snowmass 2021)

References in corpus (27)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• A Brief Introduction to PYTHIA 8.1
• A Review of Uncertainty Quantification in Deep Learning: Techniques, Applications and Challenges
• Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning
• Herwig++ Physics and Manual
• On Calibration of Modern Neural Networks
• Variational Inference with Normalizing Flows
• A tutorial on conformal prediction
• Predictive Uncertainty Estimation via Prior Networks
• Deep Gaussian Processes
• A Survey of Uncertainty in Deep Neural Networks
• 10 Simple Rules for the Care and Feeding of Scientific Data
• Fast likelihood-free cosmology with neural density estimators and active learning
• Mining gold from implicit models to improve likelihood-free inference
• Measuring Calibration in Deep Learning
• Adversarial Examples, Uncertainty, and Transfer Testing Robustness in Gaussian Process Hybrid Deep Networks
• Evaluation of probabilistic photometric redshift estimation approaches for The Rubin Observatory Legacy Survey of Space and Time (LSST)
• DisCo Fever: Robust Networks Through Distance Correlation
• Accurate emulator for the redshift-space power spectrum of dark matter halos and its application to galaxy power spectrum
• FlowPM: Distributed TensorFlow Implementation of the FastPM Cosmological N-body Solver
• The Aemulus Project IV: Emulating Halo Bias
• Differentiable Predictions for Large Scale Structure with SHAMNet
• Application of Transfer Learning to Neutrino Interaction Classification
• Machine learning synthetic spectra for probabilistic redshift estimation: SYTH-Z
• Diagnostics for Conditional Density Models and Bayesian Inference Algorithms
• Towards Instance-Wise Calibration: Local Amortized Diagnostics and Reshaping of Conditional Densities (LADaR)
• Local Calibration: Metrics and Recalibration

Cited by in corpus (4)

• Calibrating Bayesian Generative Machine Learning for Bayesiamplification
• Foundations of automatic feature extraction at LHC--point clouds and graphs
• Uncertainty Quantification From Scaling Laws in Deep Neural Networks
• Stable and Interpretable Jet Physics with IRC-Safe Equivariant Feature Extraction