Learned inference risks silent errors limited by measurement access
Predicting silent failures of learned inference under restricted access: what measurement physics and prior information explain
Machine Learning
Summary
Sometimes computer systems try to guess missing parts of data and say how sure they are, but they can be wrong without showing any warning. The authors study when these hidden errors happen and if they can be predicted before seeing the target data. They test their approach using real X-ray imaging data and find that the measurement process and decision choices explain errors more than the learned knowledge alone. Knowing this helps understand when these silent mistakes might occur.
What this means in practice
- •For medical imaging teams: Anticipate hidden errors in reconstructed X-ray CT images when access to full measurements is limited.
- •For industrial nondestructive testing teams: Improve reliability assessment of learned inference methods under restricted measurement conditions in material inspections.
Authors
Mario Gianni
Abstract
Learned priors are used to reconstruct regions that measurements barely reach, and to report uncertainty there. They can fail silently: the reported interval excludes the truth while every ground-truth-free diagnostic stays within its noise. We call such errors the blind set and ask whether they can be predicted before the target data are seen. For a correctly specified linear-Gaussian model averaged over the prior, the risk of a silent failure factorises exactly into non-detection and non-coverage, which does not hold for learned systems in general; we build a score that approximates this reference by linearising a VAE prior. In a prospective study on real X-ray CT objects (2DeteCT) under simulated restricted access, with predictions committed by hash before any trial, the score ranked the realised blindness of two learned inference systems (latent Laplace and ensemble Kalman inversion) with Spearman $ρ= 0.79$ and 0.84 over 240 pre-specified errors. Simple structural predictors would also have passed these tests. Sequential post-hoc analyses, each specified before its own computation, show that under bounded reach the learned score gave no detectable improvement over a physics-based baseline, in rankings or trial-level log scores. At the weakest attenuation studied, prior information improved rankings over the physics-based baseline, most clearly for ensemble Kalman inversion, but a Gaussian prior did as well and the gain was not detectable at the trial level. The score is not a calibrated probability. Silent failures can be anticipated here, but their structure follows mainly from measurement access and the decision quantity rather than from the learned prior.