Machine learning and Bayesian inference in high-dimensional time series
The last decade has witnessed a phenomenal increase in the application of model-based analysis techniques for the examination of multivariate time series. One particularly challenging example can be found in neuroimaging datasets. Because these datasets are extremely sparse, high-dimensional, and noisy, their analysis poses a number of compelling challenges for statistical inference and machine learning.
My research at ETH Zurich concerns the development of Bayesian approaches to the analysis of multivariate time series and their application to high-dimensional datasets, such as those obtained by functional magnetic resonance imaging (fMRI). These techniques may one day play a critical role in dissecting spectrum disorders into physiologically well-defined subtypes. I completed my PhD at the Department of Computer Science (Machine Learning Group, Professor Joachim M Buhmann) and am currently affiliated with the Department of Information Technology and Electrical Engineering (Translational Neuromodeling Unit, Professor Klaas Enno Stephan).