From brain to deep neural networks

Aim of Tutorial

The aim of this tutorial is to provide the stepping stone for machine learning enthusiasts into the area of brain pathway modelling using innovative deep learning techniques through processing and learning from electroencephalogram (EEG). An insight into EEG generation and processing will provide the audience with a better understanding of deep network structures used to learn and detect the insightful information about the deep brain function.


Plan of Tutorial

The tutorial will start with fundamental concepts and we will show how useful information can be derived from multichannel EEG data to best describe the brain normal and abnormal function. Then, state-of-the-art EEG processing for brain generative and degenerative abnormalities will be reviewed. Many impairments in brain motor function, brain responses to stimuli, and brain connectivity are manifested in EEG data. Yet, these changes are not so visible in EEG but can be cleverly be discovered using EEG processing and machine learning. This paves the way to the second part of the tutorial, where we will provide a case study on how this is achieved using deep learning.

The second part of the tutorial will take a more hands-on approach to deep learning and EEG processing with MATLAB coding. In order to facilitate simulation, some propriety anonymised EEG dataset will also be released to tutorial attendees so that they can apply the novel asymmetric auto-encoders to generate a super-resolution EEG reconstruction. This generated super-resolution EEG will then be used in a convolutional neural network to exploit temporal dependencies and LSTM to take advantage of inter-trial dependencies.


Outline of Tutorial

Foundations of EEG generation, modelling, processing, and learning

Brain connectivity and cooperative EEG learning

Collaborative ensemble learning for robustness against inter-patient variability

Asymmetric Auto-Encoders for super-resolution EEG reconstruction

Exploiting temporal dependencies and inter-trial dependencies in deep learning


Materials

The materials included in this tutorial will be in electronic format (no hard copies) and be comprised of three parts:

  1. Lecture notes

  2. MATLAB code

  3. EEG data (to be released prior to tutorial - you can request this dataset by emailing Clive Cheong Took, provided you have proof of tutorial registration and its payment).


Registration

You can register here.


Organisers

SAEID SANEI

Nottingham Trent University

Email: saeid.sanei@ntu.ac.uk

Personal Website


CLIVE CHEONG TOOK

Royal Holloway, University of London

Email: clive.cheongtook@rhul.ac.uk

Personal Website