analiza aiqbench
This paper proposes a classification variant of pairwise difference learning, where a model predicts outcome differences between instance pairs and averages over training examples, plus a way to split predictive uncertainty into aleatoric and epistemic parts. The main contribution is methodological: it adapts existing uncertainty decomposition measures to this setting, with the right choice depending on how much uncertainty the base learner itself represents. Readers should weigh the reported large-scale results on both accuracy and uncertainty calibration, since the practical value depends on how well the decomposition holds up across datasets and base learners.
Uncertainty Quantification in Pairwise Difference Learning for Classification
arXiv:2610.21554v1 Announce Type: cross Abstract: Instead of learning a mapping from instances to outcomes in the standard way, the key idea of pairwise difference learning (PDL) is to learn a function that takes two instances as input and predicts the difference between the respective outcomes. Given a function of this kind, predictions for a query instance are derived from every training example and then averaged. We propose a version of PDL for the task of classification and propose a suitable meta-learning technique for this purpose. Moreover, we address the problem of uncertainty quantification, i.e., of representing and quantifying the uncertainty of a PDL classifier's predictions in a faithful manner. Building on recent work on the decomposition of total uncertainty into aleatoric and epistemic components, we propose different ways of extending corresponding measures to the setting of PDL, depending on how and to what extent uncertainty is represented by the underlying base learner. In large-scale empirical studies, we analyze both the predictive accuracy and the uncertainty awareness of our PDL classifier.
Abstract:Instead of learning a mapping from instances to outcomes in the standard way, the key idea of pairwise difference learning (PDL) is to learn a function that takes two instances as input and predicts the difference between the respective outcomes. Given a function of this kind, predictions for a query instance are derived from every training example and then averaged. We propose a version of PDL for the task of classification and propose a suitable meta-learning technique for this purpose. Moreover, we address the problem of uncertainty quantification, i.e., of representing and quantifying the uncertainty of a PDL classifier's predictions in a faithful manner. Building on recent work on the decomposition of total uncertainty into aleatoric and epistemic components, we propose different ways of extending corresponding measures to the setting of PDL, depending on how and to what extent uncertainty is represented by the underlying base learner. In large-scale empirical studies, we analyze both the predictive accuracy and the uncertainty awareness of our PDL classifier.
32 pages. Preprint (version submitted before peer review) of the article published in Machine Learning 115, 12 (2026)
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2610.21554 [cs.LG]
(or arXiv:2610.21554v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.21554 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
https://doi.org/10.1007/s10994-025-06949-0 Focus to learn more DOI(s) linking to related resources
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 290 cuvinte
Subiecte conexe
Base Models Can Detect Hidden Intentions that Prompted Classifiers Miss
LessWrong · 12 oct. 2026
Good Practice Guide for quantifying uncertainties for machine learning models applied to photoplethysmography signals
arXiv · 23 iul. 2026
Uncertainty quantification for trustworthy deep learning: Methods and measures
arXiv · 31 iul. 2026
Impact of Data Augmentation on Confidence Calibration in Melanoma Classification
arXiv · 7 oct. 2026