Model-Free Kernel Conformal Depth Measures Algorithm for Uncertainty Quantification in Regression Models in Separable Hilbert Spaces

TL;DR

This paper introduces a model-free, kernel-based uncertainty quantification method for regression in Hilbert spaces, providing prediction regions with finite-sample guarantees and demonstrating effectiveness in simulations and health data applications.

Contribution

It proposes a novel, kernel-based, model-free algorithm for uncertainty quantification in regression within Hilbert spaces, including conformal prediction variants with finite-sample guarantees.

Findings

Faster convergence rates achieved with kernel mean embeddings.

Finite-sample prediction regions with non-asymptotic guarantees.

Effective performance demonstrated in simulations and health data applications.

Abstract

Depth measures are powerful tools for defining level sets in emerging, non--standard, and complex random objects such as high-dimensional multivariate data, functional data, and random graphs. Despite their favorable theoretical properties, the integration of depth measures into regression modeling to provide prediction regions remains a largely underexplored area of research. To address this gap, we propose a novel, model-free uncertainty quantification algorithm based on conditional depth measures--specifically, conditional kernel mean embeddings and an integrated depth measure. These new algorithms can be used to define prediction and tolerance regions when predictors and responses are defined in separable Hilbert spaces. The use of kernel mean embeddings ensures faster convergence rates in prediction region estimation. To enhance the practical utility of the algorithms with finite samples, we also introduce a conformal prediction variant that provides marginal, non-asymptotic guarantees for the derived prediction regions. Additionally, we establish both conditional and unconditional consistency results, as well as fast convergence rates in certain homoscedastic settings. We evaluate the finite--sample performance of our model in extensive simulation studies involving various types of functional data and traditional Euclidean scenarios. Finally, we demonstrate the practical relevance of our approach through a digital health application related to physical activity, aiming to provide personalized recommendations