Show simple item record

dc.contributor.authorŞensoy, Murat
dc.contributor.authorKaplan, L.
dc.contributor.authorKandemir, M.
dc.date.accessioned2020-05-22T12:14:46Z
dc.date.available2020-05-22T12:14:46Z
dc.date.issued2018
dc.identifier.issn1049-5258
dc.identifier.urihttp://hdl.handle.net/10679/6582
dc.identifier.urihttps://papers.nips.cc/paper/7580-evidential-deep-learning-to-quantify-classification-uncertainty
dc.description.abstractDeterministic neural nets have been shown to learn effective predictors on a wide range of machine learning problems. However, as the standard approach is to train the network to minimize a prediction loss, the resultant model remains ignorant to its prediction confidence. Orthogonally to Bayesian neural nets that indirectly infer prediction uncertainty through weight uncertainties, we propose explicit modeling of the same using the theory of subjective logic. By placing a Dirichlet distribution on the class probabilities, we treat predictions of a neural net as subjective opinions and learn the function that collects the evidence leading to these opinions by a deterministic neural net from data. The resultant predictor for a multi-class classification problem is another Dirichlet distribution whose parameters are set by the continuous output of a neural net. We provide a preliminary analysis on how the peculiarities of our new loss function drive improved uncertainty estimation. We observe that our method achieves unprecedented success on detection of out-of-distribution queries and endurance against adversarial perturbations.en_US
dc.description.sponsorshipUnited States Department of Defense US Army Research Laboratory (ARL) ; U.K. Ministry of Defence
dc.language.isoengen_US
dc.publisherNeural Information Processing Systems Foundationen_US
dc.relation.ispartofAdvances in Neural Information Processing Systems
dc.rightsopenAccess
dc.titleEvidential deep learning to quantify classification uncertaintyen_US
dc.typeConference paperen_US
dc.publicationstatusPublisheden_US
dc.contributor.departmentÖzyeğin University
dc.contributor.authorID(ORCID & YÖK ID 41438) Şensoy, Murat
dc.contributor.ozuauthorŞensoy, Murat
dc.identifier.volume31en_US
dc.identifier.startpage3179en_US
dc.identifier.endpage3189en_US
dc.identifier.wosWOS:000461823303020
dc.identifier.scopusSCOPUS:2-s2.0-85064805181
dc.contributor.authorMale1
dc.relation.publicationcategoryConference Paper - International - Institution Academic Staff


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record


Share this page