Informed regularization Aiding the identification of spurious correlations

dc.contributor.authorMöller, Fredrik
dc.contributor.departmentChalmers tekniska högskola / Institutionen för data och informationstekniksv
dc.contributor.examinerJohansson, Richard
dc.contributor.supervisorJohansson, Richard
dc.date.accessioned2021-03-08T07:12:59Z
dc.date.available2021-03-08T07:12:59Z
dc.date.issued2021sv
dc.date.submitted2020
dc.description.abstractToday, end-to-end neural networks that feature deep and complex architectures, are common tools to use in natural language processing. By using these methods it has become harder to identify which inputs have contributed the most to a model’s classification. This issue leads to the problem of models overfitting on features that cannot directly be identified by a developer. To open up the black box of complex deep learning natural language processing systems, this study aims to investigate what information can be extracted from the data used to train a model and how the model’s inputs are weighted during pre diction. This thesis aims to present methods that can aid in the identification of differences between the population a developer intends to model with a data set and what correlations a model makes from the true content of the data. By presenting three novel methods that can aid a developer with the task of identify ing spurious correlations, it was possible to present information regarding a spurious correlation between two pre-selected keywords and a model’s classification. It was also shown that the identification and reduction of spurious correlations is a tricky subject. Results showed that, from the reduction of the spurious correlation asso ciated with the selected keyword, the model made another correlation which could be considered as spurious.sv
dc.identifier.coursecodeMPSYSsv
dc.identifier.urihttps://hdl.handle.net/20.500.12380/302254
dc.language.isoengsv
dc.setspec.uppsokTechnology
dc.subjectNLPsv
dc.subjectExplainabilitysv
dc.subjectRegularizationsv
dc.subjectLayer-wise relevance propagationsv
dc.subjectTF-IDFsv
dc.subjectNCOFsv
dc.titleInformed regularization Aiding the identification of spurious correlationssv
dc.type.degreeExamensarbete för masterexamensv
dc.type.uppsokH
Ladda ner
Original bundle
Visar 1 - 1 av 1
Bild (thumbnail)
Namn:
CSE 21-06 Möller.pdf
Storlek:
1.88 MB
Format:
Adobe Portable Document Format
Beskrivning:
License bundle
Visar 1 - 1 av 1
Bild saknas
Namn:
license.txt
Storlek:
1.14 KB
Format:
Item-specific license agreed upon to submission
Beskrivning: