Vis enkel innførsel

dc.contributor.advisorKampffmeyer, Michael C.
dc.contributor.advisorChakraborty, Rwiddhi
dc.contributor.authorSletten, Adrian Henrik de Sena
dc.date.accessioned2024-08-23T10:32:28Z
dc.date.available2024-08-23T10:32:28Z
dc.date.issued2023-08-22
dc.description.abstractShortcut learning, the tendency for models to rely on spurious correlations, is a widespread issue in deep learning. Although being a known issue, uncovering the shortcuts present in a dataset can be a difficult task. Over the last few years, explainability methods have been leveraged to find previously unknown shortcuts within mainstream datasets. However, how to best mitigate a model’s reliance on shortcuts, is not a matter that is widely agreed on. Recently, the concept of group robustness has appeared as a potential way for mitigating shortcuts. In group robustness, the data in each class is divided into subclasses where some contain the shortcut features while some other subclass does not. By optimising a model to increase the worst group performance, the model learns to perform well across groups, mitigating reliance on shortcuts. One notable limitation that exists for current group robustness methods is their reliance on group labels to guarantee performance improvements. The issue with this is that acquiring these additional labels is a difficult and timeconsuming task. Therefore we propose, eXplainability-based Feature Reweighting (XFR), a group-unsupervised group robustness method. Our proposed method leverages the clustering of explainability heatmaps to estimate pseudolabels for groups in a dataset and afterwards uses these labels to improve group robustness. In our results, we show that XFR clearly improves group robustness compared to standardly trained models (ERM). We also show that performance is on par with, and sometimes even surpasses, methods that partially or fully utilise group labels.en_US
dc.identifier.urihttps://hdl.handle.net/10037/34375
dc.language.isoengen_US
dc.publisherUiT Norges arktiske universiteten_US
dc.publisherUiT The Arctic University of Norwayen_US
dc.rights.accessRightsopenAccessen_US
dc.rights.holderCopyright 2023 The Author(s)
dc.rights.urihttps://creativecommons.org/licenses/by-nc-sa/4.0en_US
dc.rightsAttribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)en_US
dc.subject.courseIDFYS-3941
dc.subjectVDP::Technology: 500::Information and communication technology: 550::Computer technology: 551en_US
dc.subjectVDP::Teknologi: 500::Informasjons- og kommunikasjonsteknologi: 550::Datateknologi: 551en_US
dc.subjectVDP::Technology: 500::Information and communication technology: 550::Other information technology: 559en_US
dc.subjectVDP::Teknologi: 500::Informasjons- og kommunikasjonsteknologi: 550::Annen informasjonsteknologi: 559en_US
dc.subjectVDP::Mathematics and natural science: 400::Physics: 430en_US
dc.subjectVDP::Matematikk og Naturvitenskap: 400::Fysikk: 430en_US
dc.subjectVDP::Mathematics and natural science: 400::Information and communication science: 420::Simulation, visualization, signal processing, image processing: 429en_US
dc.subjectVDP::Matematikk og Naturvitenskap: 400::Informasjons- og kommunikasjonsvitenskap: 420::Simulering, visualisering, signalbehandling, bildeanalyse: 429en_US
dc.subjectVDP::Mathematics and natural science: 400::Mathematics: 410::Applied mathematics: 413en_US
dc.subjectVDP::Matematikk og Naturvitenskap: 400::Matematikk: 410::Anvendt matematikk: 413en_US
dc.subjectVDP::Technology: 500::Information and communication technology: 550::Computer technology: 551en_US
dc.subjectVDP::Teknologi: 500::Informasjons- og kommunikasjonsteknologi: 550::Datateknologi: 551en_US
dc.titleLeveraging Explainability Maps for Group-unsupervised Robustness to Spurious Correlationsen_US
dc.typeMaster thesisen_US
dc.typeMastergradsoppgaveen_US


Tilhørende fil(er)

Thumbnail
Thumbnail

Denne innførselen finnes i følgende samling(er)

Vis enkel innførsel

Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)
Med mindre det står noe annet, er denne innførselens lisens beskrevet som Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)