analiza aiqbench
This paper extends pairwise difference learning, a meta-learning method previously proposed for regression, to classification by recasting the task as a binary problem over pairs of training examples and averaging predictions across the training set. The reported gains over existing methods come from a large-scale empirical study, so the practical value depends on how those benchmarks compare with current baselines and on the released Python package. For ML practitioners, the main question is whether the pairwise formulation offers a useful alternative to standard classifiers at a reasonable computational cost.
Pairwise Difference Learning for Classification
arXiv:2406.20031v2 Announce Type: replace-cross Abstract: Pairwise difference learning (PDL) has recently been introduced as a new meta-learning technique for regression. Instead of learning a mapping from instances to outcomes in the standard way, the key idea 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. This paper extends PDL toward the task of classification and proposes a meta-learning technique for inducing a PDL classifier by solving a suitably defined (binary) classification problem on a paired version of the original training data. We analyze the performance of the PDL classifier in a large-scale empirical study and find that it outperforms state-of-the-art methods in terms of prediction performance. Last but not least, we provide an easy-to-use and publicly available implementation of PDL in a Python package.
Abstract:Pairwise difference learning (PDL) has recently been introduced as a new meta-learning technique for regression. Instead of learning a mapping from instances to outcomes in the standard way, the key idea 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. This paper extends PDL toward the task of classification and proposes a meta-learning technique for inducing a PDL classifier by solving a suitably defined (binary) classification problem on a paired version of the original training data. We analyze the performance of the PDL classifier in a large-scale empirical study and find that it outperforms state-of-the-art methods in terms of prediction performance. Last but not least, we provide an easy-to-use and publicly available implementation of PDL in a Python package.
v2: corrected the e-mail address of E. Hüllermeier and the ORCID of M.K. Belaid; added a reference to the published version. Preprint (version submitted before peer review) of the paper published in Discovery Science (DS 2024), LNCS vol. 15244, Springer
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2406.20031 [cs.LG]
(or arXiv:2406.20031v2 [cs.LG] for this version)
Discovery Science (DS 2024), Lecture Notes in Computer Science, vol. 15244, pp. 284-299, Springer (2025)
https://doi.org/10.1007/978-3-031-78980-9_18 DOI(s) linking to related resources
Text integral sub licența deschisă a publicației · 224 cuvinte
Subiecte conexe
Base Models Can Detect Hidden Intentions that Prompted Classifiers Miss
LessWrong · 12 oct. 2026
Challenge: Hand coding weights for efficient sequence memorisation
AI Alignment Forum · 23 iul. 2026
Efficient Pairwise Difference Learning: A Comparative Study of Encoding Strategies and Training Pair Selection
arXiv · 17 oct. 2026
Uncertainty Quantification in Pairwise Difference Learning for Classification
arXiv · 17 oct. 2026