Trustworthy Prediction with Gaussian Process Knowledge Scores
stat.ML, cs.LG, eess.SP
Submitted: 2025-06-23
Updated: 2025-06-23
Comments: 6 pages, 5 figures, to be published in the Proceedings of the European Signal Processing Conference (EUSIPCO)
Journal ref: 2025 33rd European Signal Processing Conference (EUSIPCO), Palermo, Italy, 2025, pp. 875-879
DOI: 10.23919/EUSIPCO63237.2025.11226511
Code: https://github.com/KurtButler/GP-knowledge
License: http://creativecommons.org/licenses/by/4.0/
The gist: Probabilistic models are often used to make predictions in regions of the data space where no observations are available…
Terminology
Abstract
Probabilistic models are often used to make predictions in regions of the data space where no observations are available, but it is not always clear whether such predictions are well-informed by previously seen data. In this paper, we propose a knowledge score for predictions from Gaussian process regression (GPR) models that quantifies the extent to which observing data have reduced our uncertainty about a prediction. The knowledge score is interpretable and naturally bounded between 0 and 1. We demonstrate in several experiments that the knowledge score can anticipate when predictions from a GPR model are accurate, and that this anticipation improves performance in tasks such as anomaly detection, extrapolation, and missing data imputation. Source code for this project is available online at https://github.com/KurtButler/GP-knowledge.
Related papers
- Uniform-in-Time Weak Propagation-of-Chaos in Shallow Neural Networks
- Diffusion Models Observe Only Gradients: A Geometric Perspective on Score Matching Errors
- Quantifying Error Propagation and Model Collapse in Diffusion Models
- Behavior of prediction performance metrics with rare events
- Statistical Inference and Learning for Shapley Additive Explanations (SHAP)
- Control Variate Design for Score-Function Gradient Estimators in Importance-Weighted Variational Inference