National Technical University of Athens (NTUA)
Algorithmic approaches for efficiency and explainability in learning problems
Abstract
dc:descriptionThis doctoral dissertation focuses on the development of supervised and unsupervised machine learning algorithms with an emphasis on two fundamental aspects: computational efficiency and explainability. The need for efficient and explainable models is particularly crucial in high-stakes domains such as healthcare, education, and finance, where automated decisions can directly impact human lives. Within this context, the dissertation addresses three core problems. (1) In the label ranking problem, which generalizes multiclass classification, the task is modeled as a nonparametric regression problem. Interpretable algorithms based on decision trees and random forests are proposed, accompanied by new theoretical guarantees and extensions of existing statistical bounds to noisy and incomplete data. (2) For global counterfactual explanations, the problem is formulated as a multi-objective optimization balancing interpretability, cost, and effectiveness. Its computational hardness is formally established, and efficient clustering-based algorithms are introduced, achieving improved trade-offs across the competing criteria compared to existing approaches. (3) In the time series clustering problem, a two-phase algorithm is proposed that first applies Sparse Gaussian Process Regression to reduce the dimensionality of time series before clustering, significantly improving computational and memory efficiency while preserving clustering quality and enhancing interpretability. Overall, the dissertation makes both theoretical and practical contributions, providing new computational and statistical guarantees for fundamental learning problems and offering scalable, interpretable algorithms applicable to large-scale, real-world datasets in socially impactful domains.
Degree
thesis:*- Grantor dc:publisher
- National Technical University of Athens (NTUA)
- Year dc:date
- 2025
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Psaroudaki, Eleni
- Ψαρουδάκη, Ελένη
Subjects
dc:subject × 22- Θεωρητική μηχανική μάθηση
- Εξηγήσιμη τεχνητή νοημοσύνη
- Κατάταξη ετικετών
- Καθολικές επεξηγήσεις με αντιπαραδείγματα
- Ομαδοποίηση χρονοσειρών
- Theoretical machine learning
- Explainable artificial intelligence
- Label ranking
- Global counterfactual explanations
- Time-series clustering
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Φυσικές Επιστήμες
- Τεχνητή νοημοσύνη
- Επιστήμη Ηλεκτρολόγου Μηχανικού, Ηλεκτρονικού Μηχανικού, Μηχανικού Η/Υ
- Επιστήμες Μηχανικού και Τεχνολογία
- Υπολογιστές, Υλικό (hardware) και Αρχιτεκτονική
- Computer and Information Sciences
- Natural Sciences
- Artificial Intelligence
- Electrical Engineering, Electronic Engineering, Information Engineering
- Engineering and Technology
- Computer science, Hardware and Architecture
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/60882