121191 20240319081019.0 doi 10.1109/TGRS.2022.3205119 sideral 130613 ART-2022-130613 eng Alcolea, Adrián Universidad de Zaragoza (orcid)0000-0002-7057-4283 Bayesian neural networks to analyze hyperspectral datasets using uncertainty metrics 2022 Access copy available to the general public Unrestricted Machine learning techniques, and specifically neural networks, have proved to be very useful tools for image classification tasks. Nevertheless, measuring the reliability of these networks and calibrating them accurately are very complex. This is even more complex in a field like hyperspectral imaging, where labeled data are scarce and difficult to generate. Bayesian neural networks (BNNs) allow to obtain uncertainty metrics related to the data processed (aleatoric), and to the uncertainty generated by the model selected (epistemic). On this work, we will demonstrate the utility of BNNs by analyzing the uncertainty metrics obtained by a BNN over five of the most used hyperspectral images datasets. In addition, we will illustrate how these metrics can be used for several practical applications such as identifying predictions that do not reach the required level of accuracy, detecting mislabeling in the dataset, or identifying when the predictions are affected by the increase of the level of noise in the input data. info:eu-repo/grantAgreement/ES/AEI/PID2019-105660RB-C21-AEI-10.13039-501100011033 info:eu-repo/grantAgreement/ES/DGA-ESF/T58-20R info:eu-repo/semantics/openAccess All rights reserved http://www.europeana.eu/rights/rr-f/ 8.2 2022 ENGINEERING, ELECTRICAL & ELECTRONIC 27 / 274 = 0.099 2022 Q1 T1 IMAGING SCIENCE & PHOTOGRAPHIC TECHNOLOGY 5 / 28 = 0.179 2022 Q1 T1 REMOTE SENSING 4 / 34 = 0.118 2022 Q1 T1 GEOCHEMISTRY & GEOPHYSICS 5 / 87 = 0.057 2022 Q1 T1 2.4 2022 Electrical and Electronic Engineering 2022 Q1 Earth and Planetary Sciences (miscellaneous) 2022 Q1 10.9 2022 info:eu-repo/semantics/article info:eu-repo/semantics/acceptedVersion Resano, Javier Universidad de Zaragoza (orcid)0000-0002-7532-2720 5007 035 Universidad de Zaragoza Dpto. Informát.Ingenie.Sistms. Área Arquit.Tecnología Comput. 60 (2022), 5537810 [10 pp.] IEEE trans. geosci. remote sens. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING 0196-2892 13188744 http://zaguan.unizar.es/record/121191/files/texto_completo.pdf Postprint 4009682 http://zaguan.unizar.es/record/121191/files/texto_completo.jpg?subformat=icon icon Postprint oai:zaguan.unizar.es:121191 articulos driver 2024-03-18-15:58:12 ARTICLE