analiza aiqbench
This research investigates a critical deployment risk where conditional segmentation models trained on clean auxiliary signals fail when faced with noisier real-world data, specifically in echocardiographic imaging. The study quantifies this "conditioning-availability bias" and demonstrates that performance gaps persist even if mean Dice scores remain stable, highlighting the fragility of oracle-dependent inference pathways. These findings are essential for advancing robust machine learning practices in medical AI by addressing shortcut learning caused by auxiliary-variable shifts.
When Oracle Conditioning Misleads Deployment: Conditioning-Availability Bias in Echocardiographic Segmentation
arXiv:2608.03342v1 Announce Type: cross Abstract: Conditional segmentation models may be trained and evaluated with auxiliary signals cleaner than those available at deployment. We study this protocol-level manifestation of shortcut learning and auxiliary-variable shift in phase-conditioned echocardiographic segmentation. The complementary gap pair measures loss on the deployable oracle-estimated pathway and probes sensitivity on the oracle-random pathway. On held-out CAMUS data, one strong-cyclic, oracle-selected run fails severely with estimated phase, while sensitivity to incorrect phase persists across three runs. On EchoNet-Dynamic, the current estimator remains usable, but random-phase testing reveals strong latent sensitivity. Deployment-aware checkpoint selection and phase perturbation reduce both gaps with little change in mean Dice. Exploratory subgroup analyses quantify variation across measured strata, and a downstream ejection fraction (EF) audit shows that recovering segmentation does not necessarily recover EF error or signed bias. Together, the gaps test whether oracle-conditioned performance survives the inference pathway actually available at deployment.
Abstract:Conditional segmentation models may be trained and evaluated with auxiliary signals cleaner than those available at deployment. We study this protocol-level manifestation of shortcut learning and auxiliary-variable shift in phase-conditioned echocardiographic segmentation. The complementary gap pair measures loss on the deployable oracle-estimated pathway and probes sensitivity on the oracle-random pathway. On held-out CAMUS data, one strong-cyclic, oracle-selected run fails severely with estimated phase, while sensitivity to incorrect phase persists across three runs. On EchoNet-Dynamic, the current estimator remains usable, but random-phase testing reveals strong latent sensitivity. Deployment-aware checkpoint selection and phase perturbation reduce both gaps with little change in mean Dice. Exploratory subgroup analyses quantify variation across measured strata, and a downstream ejection fraction (EF) audit shows that recovering segmentation does not necessarily recover EF error or signed bias. Together, the gaps test whether oracle-conditioned performance survives the inference pathway actually available at deployment.
Accepted for publication in the MICCAI 2026 Workshop on Fairness of AI in Medical Imaging (FAIMI 2026). To appear in Springer Lecture Notes in Computer Science (LNCS)
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2608.03342 [cs.CV]
(or arXiv:2608.03342v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2608.03342 arXiv-issued DOI via DataCite (pending registration)
From: Dang Cao Pham Minh Mr. [view email] [v1] Tue, 4 Aug 2026 08:52:43 UTC (255 KB)
Text integral sub licența deschisă a publicației · 212 cuvinte
Subiecte conexe
Does ChatGPT really have a strong left-wing bias?
LessWrong · 27 iul. 2026
J++ Lens: Jacobian Filtering Enables More Faithful Workspace Lenses
LessWrong · 8 oct. 2026
Why do models task game?
AI Alignment Forum · 6 aug. 2026
Ranked Right, Sized Wrong: Ensemble Spread Fails Silently at Streamflow Extremes
arXiv · 16 oct. 2026