Self-Supervised Conformal Prediction with Equivariant Bootstrapping for Image Uncertainty Quantification

TL;DR

This paper presents a self-supervised conformal prediction method using equivariant bootstrapping to quantify uncertainty in image inverse problems without requiring ground truth calibration data.

Contribution

It introduces a novel uncertainty quantification approach combining equivariant bootstrapping and conformal prediction, avoiding the need for ground truth calibration data.

Findings

Effective uncertainty quantification demonstrated in weak lensing mass-mapping.

Method reduces biases caused by reliance on specific calibration data.

Provides heuristic coverage guarantees through equivariant bootstrapping.

Abstract

Inverse problems are ubiquitous in modern scientific studies and involve recovering an underlying signal from noisy observations often transformed by a measurement operator. These problems are frequently ill-posed, particularly in imaging, leading to multiple plausible solutions and considerable uncertainty in reconstructed images. In fields like the physical and biological sciences, accurate uncertainty quantification (UQ) is critical for trustworthy scientific analyses and confident diagnoses. Current UQ methods for imaging often fall short; they can be inaccurate, or require unavailable or difficult-to-acquire ground truth data for calibration, which can introduce hidden biases due to distribution shifts between calibration and observed data. We introduce a UQ approach that leverages equivariant bootstrapping to generate heuristic coverages by exploiting data symmetries. We then refine these coverages through a conformal prediction calibration step, while crucially employing a self-supervised approach to avoid the need for ground truth calibration data. We demonstrate this method with weak lensing mass-mapping, where we aim to reconstruct the convergence field from shear measurements of distant galaxies weakly-lensed by gravitational fields. Mass-mapping in particular benefits from the self-supervised approach, as simulating calibration data is expensive and relies on specific cosmological models that could introduce biases in downstream cosmological inference tasks.