analiza aiqbench
This arXiv paper examines active learning for new classes, where an anomaly detector ranks unfamiliar examples for expert labelling. Across five benchmarks, confirming the top-ranked anomalies taught classifiers less than confirming random examples, and replacing nearby known classes with distant ones cut the confirmations needed. For ML practitioners, the takeaway is that anomaly scores are useful for discovering new classes but can be a poor guide for selecting training examples that define them, so sampling strategy needs separate evaluation.
Anomaly-ranked confirmation slows new-class learning
arXiv:2611.03250v1 Announce Type: cross Abstract: Anomaly detectors rank unfamiliar examples for expert confirmation, and the confirmed examples teach a classifier to recognise a new class. Across five benchmarks from physics, medicine and computer vision, the members flagged by the detector are those the classifier learns first. In three systems, confirming the highest-ranked members of a new class gives lower recognition than confirming the same number of random members. Handing subsequent selection to the promoted classifier recovers part of this loss in two systems. A paired intervention further shows that replacing nearby known classes with distant ones sharply reduces the confirmations needed for recognition. These results distinguish finding a new class from selecting examples that teach its extent: within the tested classes, concentrating confirmations at the top of the anomaly ranking makes learning less efficient.
Abstract:Anomaly detectors rank unfamiliar examples for expert confirmation, and the confirmed examples teach a classifier to recognise a new class. Across five benchmarks from physics, medicine and computer vision, the members flagged by the detector are those the classifier learns first. In three systems, confirming the highest-ranked members of a new class gives lower recognition than confirming the same number of random members. Handing subsequent selection to the promoted classifier recovers part of this loss in two systems. A paired intervention further shows that replacing nearby known classes with distant ones sharply reduces the confirmations needed for recognition. These results distinguish finding a new class from selecting examples that teach its extent: within the tested classes, concentrating confirmations at the top of the anomaly ranking makes learning less efficient.
Instrumentation and Methods for Astrophysics (astro-ph.IM); Machine Learning (cs.LG); Data Analysis, Statistics and Probability (physics.data-an)
arXiv:2611.03250 [astro-ph.IM]
(or arXiv:2611.03250v1 [astro-ph.IM] for this version)
https://doi.org/10.48550/arXiv.2611.03250 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
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 · 235 cuvinte