Analyzing Connectivity between MEG Channels using PLV

This is a short update to my earlier post on MEG data classification. As a short recap, I managed to get data classification accuracy up to 60-65% - but some of the classes were classified almost systematically incorrectly while two of the classes were categorized almost flawlessly. The classifier was based on classification of interaction … Continue reading Analyzing Connectivity between MEG Channels using PLV

MEG Signal Classification using PLV and Neural Networks

Despite I've written two posts on modelling MEG signals, I haven't yet written a single post on my attempts to classify the signals. I lately experimented with a new approach and got some results that seemed to be worth documenting. The high-level idea is to first determine which brain regions were interacting and pass this information … Continue reading MEG Signal Classification using PLV and Neural Networks

MEG Signal Modeling using Recurrent Neural Networks

This is a short update to my earlier post from 2015 on modeling MEG signals using CRBMs. CRBMs were outdated already at the time I wrote the post and one of the development areas was to try out recurrent neural networks (RNNs). I'm describing here the results of my weekend project on applying RNNs to time series prediction of MEG data. … Continue reading MEG Signal Modeling using Recurrent Neural Networks

Modeling MEG signals using CRBMs

Machine learning appears to be a hot topic nowadays - which is a good excuse for me to write my first post about Conditional Restricted Boltzmann Machines (CRBMs). I gather here my results on applying CRBMs to magnetoencephalographic (MEG) signals to create a generative model that produces oscillations which resemble the original data. Don't expect anything too scientific from this … Continue reading Modeling MEG signals using CRBMs