Summary
Arinbjörn Kolbeinsson is an honorary research associate in the Department of Epidemiology and Biostatistics and a visiting scholar at the University of Virginia.
His work focuses on developing machine learning methods that learn from high-dimensional biomedical data including genetic, brain MRI and hospital diagnosis data for personalised health outcome prediction.
Recent research activities include an interpretable deep learning system that learns a biomarker from brain structure that associates with both lifestyle factors, risk factors and diseases. In another a novel prediction model for polygenic scores.
Arinbjörn completed his PhD in 2021 at Imperial College London, advised by Professor Ioanna Tzoulaki, Dr Sarah Filippi, Dr Yannis Panagakis and Professor Paul Elliott.
During his PhD, Arinbjörn completed a research internship at Samsung AI and developed tensor methods for improved robustness of deep learning methods in collaboration with Dr Jean Kossaifi and Professor Anima Anandkumar at Nvidia/Caltech.
Publications
Journals
Lightley J, Gorlitz F, Kumar S, et al., 2021, Robust deep learning optical autofocus system applied to automated multiwell plate single molecule localization microscopy, Journal of Microscopy, ISSN:0022-2720
Kolbeinsson A, Kossaifi J, Panagakis I, et al., 2021, Tensor dropout for robust learning, Ieee Journal of Selected Topics in Signal Processing, Vol:15, ISSN:1932-4553, Pages:630-640
Kolbeinsson A, Filippi S, Panagakis I, et al., 2020, Accelerated MRI-predicted brain ageing and its associations with cardiometabolic and brain disorders, Scientific Reports, Vol:10, ISSN:2045-2322
Bintsi KM, Baltatzis V, Kolbeinsson A, et al., 2020, Patch-Based Brain Age Estimation from MR Images, Lecture Notes in Computer Science (including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol:12449 LNCS, ISSN:0302-9743, Pages:98-107