direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Inhalt des Dokuments

Master Thesis: Implementation and Evaluation of Distributed Online Clustering Algorithm

Implementation and Evaluation of Distributed Online Clustering Algorithm


Recent improvements in hardware technology made possible to process immense amount of data. This opened new possibilities to analyze information and apply computationally expensive machine learning algorithms to real world problems. Clustering is a machine-learning category that is significantly influenced by this transition. The need to process large-scale data gave rise to the new clustering branch of online clustering algorithms. These algorithms enable processing large datasets and data streams. Hence, it is now possible to create real-time applications that are based on online clustering methods. Such method is the BIRCH clustering algorithm, which is designed to handle large datasets and enables real-time data processing at a very high speed. Similar to other clustering methods, this method requires some degree of knowledge about the dataset. As a consequence, the user needs to provide a total number of desired clusters or input threshold parameter that governs the cluster creation. This thesis addresses the issue of the BIRCH input parameter, demonstrating that it is possible to set the BIRCH threshold automatically. My experimental results have confirmed that is possible to design a fully automatized clustering algorithm that does not require the user assistance or prior knowledge. Reviewing the relevant literature on online clustering methods, this thesis also presents implementation of methods needed to derive the automatic BIRCH threshold.


Supervisor: Ana Kosareva, Boris Lorbeer

Type:  Master Thesis

Duration: 6 months

Zusatzinformationen / Extras

Quick Access:

Schnellnavigation zur Seite über Nummerneingabe

TU Berlin - Service-centric Networking - TEL 19
Ernst-Reuter-Platz 7
10587 Berlin, Germany
Phone: +49 30 8353 58811
Fax: +49 30 8353 58409