Graph neural networks for optimal and efficient generation of textual counterfactuals

Lymperopoulos, Dimitris; Λυμπερόπουλος, Δημήτρης

dc.contributor.author	Lymperopoulos, Dimitris	en
dc.contributor.author	Λυμπερόπουλος, Δημήτρης	el
dc.date.accessioned	2025-03-28T09:22:09Z
dc.date.available	2025-03-28T09:22:09Z
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/61523
dc.identifier.uri	http://dx.doi.org/10.26240/heal.ntua.29219
dc.rights	Default License
dc.subject	Νευρωνικά Δίκτυα Γράφων	el
dc.subject	Εξηγήσεις με αντιπαράδειγμα	el
dc.subject	Διγράφοι	el
dc.subject	Ορθογώνιο Πρόβλημα Γραμμικής Ανάθεσης	el
dc.subject	Τεχνητή νοημοσύνη	el
dc.subject	Graph neural networks	en
dc.subject	Counterfactual Explanations	en
dc.subject	Bipartite graphs	en
dc.subject	Rectangular Linear Assignment Problem	en
dc.subject	Artificial Intelligence	en
dc.title	Graph neural networks for optimal and efficient generation of textual counterfactuals	en
heal.type	bachelorThesis
heal.classification	Counterfactual Explanations	en
heal.language	el
heal.language	en
heal.access	free
heal.recordProvider	ntua	el
heal.publicationDate	2024-10-24
heal.abstract	Counterfactual explanations provide reasoning in the form of changes needed to be made in order for a model to make a different decision. When the model in question is a black-box classifier and the input consists of textual data, many counterfactual editors attempt to gain insights about the inner workings of the model by slightly altering the original instances. However most of them are computationally expensive due to the massive space of alternatives one has to search when altering a text. In this thesis, we propose using the recently thriving deep learning models which specifically operate on graph structured data, called Graph Neural Networks (GNN). We present an editor that generates semantically edited inputs, known as counterfactual interventions, which change the model prediction, thus providing a form of counterfactual explanations for the model. The editor utilizes a special graph type knows as a bipartite graph (or bigraph) along with a GNN that we developed so that it simulates the solution to the Rectangular Linear Assignment Problem (RLAP). During our experiments, we showcase the editor’s flexible nature, and discuss multiple trade-offs regarding explainability, minimality and speed. We test our editor on two NLP tasks - binary sentiment classification and topic classification - and show that the generated edits are contrastive, fluent and minimal, while the whole process remains significantly faster than other state-of-the-art counterfactual editors.	en
heal.advisorName	Stamou, Giorgos	en
heal.committeeMemberName	Stamou, Giorgos	en
heal.committeeMemberName	Voulodimos, Athanasios	en
heal.committeeMemberName	Kollias, Stefanos	en
heal.academicPublisher	Εθνικό Μετσόβιο Πολυτεχνείο. Σχολή Ηλεκτρολόγων Μηχανικών και Μηχανικών Υπολογιστών. Τομέας Τεχνολογίας Πληροφορικής και Υπολογιστών	el
heal.academicPublisherID	ntua
heal.numberOfPages	102 σ.	el
heal.fullTextAvailability	false