Aggregating Image Segmentation Predictions with Probabilistic Risk Control Guarantees

In this work, we introduce a framework to combine arbitrary image segmentation algorithms from different agents under data privacy constraints to produce an aggregated prediction set satisfying finite-sample risk control guarantees. We leverage distribution-free uncertainty quantification techniques...

Full description

Saved in:
Bibliographic Details
Main Authors: Joaquin Alvarez, Edgar Roman-Rangel
Format: Article
Language:English
Published: MDPI AG 2025-05-01
Series:Mathematics
Subjects:
Online Access:https://www.mdpi.com/2227-7390/13/11/1711
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849330965630943232
author Joaquin Alvarez
Edgar Roman-Rangel
author_facet Joaquin Alvarez
Edgar Roman-Rangel
author_sort Joaquin Alvarez
collection DOAJ
description In this work, we introduce a framework to combine arbitrary image segmentation algorithms from different agents under data privacy constraints to produce an aggregated prediction set satisfying finite-sample risk control guarantees. We leverage distribution-free uncertainty quantification techniques in order to aggregate deep neural networks for image segmentation tasks. Our method can be applied in settings to merge the predictions of multiple agents with arbitrarily dependent prediction sets. Moreover, we perform experiments in medical imaging tasks to illustrate our proposed framework. Our results show that the framework reduced the empirical false positive rate by 50% without compromising the false negative rate, with respect to the false positive rate of any of the constituent models in the aggregated prediction algorithm.
format Article
id doaj-art-93a3d7f4e6c04d2988993cb78deb7661
institution Kabale University
issn 2227-7390
language English
publishDate 2025-05-01
publisher MDPI AG
record_format Article
series Mathematics
spelling doaj-art-93a3d7f4e6c04d2988993cb78deb76612025-08-20T03:46:46ZengMDPI AGMathematics2227-73902025-05-011311171110.3390/math13111711Aggregating Image Segmentation Predictions with Probabilistic Risk Control GuaranteesJoaquin Alvarez0Edgar Roman-Rangel1Department of Computer Science, Instituto Tecnológico Autónomo de México, Mexico City 01080, MexicoDepartment of Computer Science, Instituto Tecnológico Autónomo de México, Mexico City 01080, MexicoIn this work, we introduce a framework to combine arbitrary image segmentation algorithms from different agents under data privacy constraints to produce an aggregated prediction set satisfying finite-sample risk control guarantees. We leverage distribution-free uncertainty quantification techniques in order to aggregate deep neural networks for image segmentation tasks. Our method can be applied in settings to merge the predictions of multiple agents with arbitrarily dependent prediction sets. Moreover, we perform experiments in medical imaging tasks to illustrate our proposed framework. Our results show that the framework reduced the empirical false positive rate by 50% without compromising the false negative rate, with respect to the false positive rate of any of the constituent models in the aggregated prediction algorithm.https://www.mdpi.com/2227-7390/13/11/1711risk controlguaranteesdistribution-freeuncertainty quantificationensemble learningpolyps
spellingShingle Joaquin Alvarez
Edgar Roman-Rangel
Aggregating Image Segmentation Predictions with Probabilistic Risk Control Guarantees
Mathematics
risk control
guarantees
distribution-free
uncertainty quantification
ensemble learning
polyps
title Aggregating Image Segmentation Predictions with Probabilistic Risk Control Guarantees
title_full Aggregating Image Segmentation Predictions with Probabilistic Risk Control Guarantees
title_fullStr Aggregating Image Segmentation Predictions with Probabilistic Risk Control Guarantees
title_full_unstemmed Aggregating Image Segmentation Predictions with Probabilistic Risk Control Guarantees
title_short Aggregating Image Segmentation Predictions with Probabilistic Risk Control Guarantees
title_sort aggregating image segmentation predictions with probabilistic risk control guarantees
topic risk control
guarantees
distribution-free
uncertainty quantification
ensemble learning
polyps
url https://www.mdpi.com/2227-7390/13/11/1711
work_keys_str_mv AT joaquinalvarez aggregatingimagesegmentationpredictionswithprobabilisticriskcontrolguarantees
AT edgarromanrangel aggregatingimagesegmentationpredictionswithprobabilisticriskcontrolguarantees