Data clustering under the prism of AutoML
Συσταδοποίηση δεδομένων υπό το πρίσμα τεχνικών αυτόματης μηχανικής μάθησης
View/
Keywords
Machine learning ; Clustering ; Hyperparameter optimization ; Meta-learningAbstract
The training of machine learning models has experienced rapid growth in recent years. However, due to the large number of available algorithms and their dependency on hyperparameters, identifying the optimal model remains a challenging and time-consuming process. While numerous Automated Machine Learning (AutoML) systems have been developed for supervised learning problems, tasks that belong to unsupervised learning, such as clustering, have not received the same level of attention. This thesis presents the challenges associated with automating clustering, as well as how state-of-the-art methods, many of which have been successfully applied in supervised learning, can be leveraged to develop AutoML systems for clustering. Furthermore, we review existing research frameworks and the diverse approaches proposed for addressing this problem. Finally, we propose a new methodology for designing an AutoML system for clustering, implement it experimentally and analyze its results.