Aarhus University Seal

GPU-accelerated unsupervised machine learning on modern desktop hardware

PhD defence, Thursday 22 September 2022. Jakob Rødsgaard Jørgensen.

Jakob Rødsgaard Jørgensen

During his Ph.D. studies, Jakob Rødsgaard Jørgensen researched GPU-accelerating algorithms for data mining, also known as unsupervised machine learning. Contrary to most supervised machine learning, data mining algorithms have not received the same attempts of GPU acceleration. More specifically, Jakob Rødsgaard Jørgensen study how clustering, subspace clustering, and projected clustering algorithms can be adapted to fit the model of computation of the GPU.
The research has resulted in several new GPU-parallelized adaptations of these. Besides achieving speedup by GPU-parallelization, the algorithms utilize different strategies for reducing the runtime of the range queries dependent on the context of each algorithm.

These strategies include pruning, summarization, and using indexing structures.

The PhD study was completed at Department of Computer Science, Faculty of Natural Sciences, Aarhus University.

This summary was prepared by the PhD student.

Time: Thursday, 22 September 2022 at 13:00
Place: Building 5342, room 333, Lecture Theatre Ada-333, Department of Computer Science, Aarhus University, Åbogade 34, Aarhus N
Title of PhD thesis: Parallel algorithms for clustering, subspace clustering, and projected clustering on the GPU

Contact information: Jakob Rødsgaard Jørgensen, e-mail: jakobrj@cs.au.dk, tel.: +45 27640053

Members of the assessment committee:

Associate Professor Rafael Sachetto Oliveira, Department of Computer Science, Federal University of São João del-Rei, Brazil

Professor Christian Böhm, Department of Comptuer Science, University of Munich, Germany

Associate Professor Peyman Afshani (chair), Department of Computer Science, Aarhus University, Denmark

Main supervisor:

Professor Ira Assent, Department of Computer Science, Aarhus University, Denmark

Language: The PhD dissertation will be defended in English

The defence is public.
The PhD thesis is available for reading at the Graduate School of Natural Sciences/GSNS,

Jens Baggesens Vej 53, building 5221, 8200 Aarhus N.

16882 / i43