I received a diploma degree in computer science in 2009 from HU Berlin, and afterwards volunteered as a Human Rights Observer for IPON in the Philippines. From 2010 to 2015, I then pursued a PhD and worked as a teaching and research assistant in the Machine Learning Group at TU Berlin. Since 2016, I'm working as a Senior Software Engineer at IVU Traffic Technologies.
Multiple Artifact Rejection Algorithm (MARA)
An important pre-processing step for the analysis of brain signals acquired with the electroencephalogram (EEG) is the removal of artifacts, caused by e.g. eye and muscle movements. A common approach is to linearly decompose the EEG signals into source components using Independent Component Analysis (ICA). Artifactual components then have to be identified and discarded. MARA is an open-source EEGLAB plug-in which solves the binary classification problem of labeling independent source components into artifactual and non-artifactual components.
Download the plug-in here.
Time-Reversed Granger causality (TRGC)
In time-series analysis, inference about cause-effect relationships is commonly based on the principle that the cause should precede its effect, using variants of the Granger causality methodology. One problem (out of many) is that spurious Granger causality can occur due to measurement noise. To address this issue, Haufe et al. (2013) suggested to contrast causality scores obtained on the original time series to those obtained on time-reversed signals. The intuitive idea behind this approach is that, if temporal order is crucial to tell a cause from an effect, directed information flow should be reduced if the temporal order is reversed.
Theoretical guarantees and code that generated accompanying simulations can be found here.
Excercice Machine Learning 2: SS 2011
Seminar Classical Topics in Machine Learning: WS 2010/2011
Seminar Kausalanalyse: WS 2012/2013