Hands-on Brain-Computer Interfacing Bootcamp: — Introduction

Saturdays.AI is a non-profit global initiative to democratize access to Artificial Intelligence education at the quality and rigour of the world’s best universities, in the form of 14-week practice bootcamps beyond the conventional education path

We’ve partnered with Bitbrain, a leading world-class neurotechnology company, to explore projects at the intersection of Brain-Computer Interfacing (BCI) and Artificial Intelligence (AI).

Weeks: 0 ・ 1 ・ 2 ・ 3 ・ 4 ・ 5 ・ 6 ・ 7 ・ 8 ・ 9 ・ 10 ・ 11 ・ 12 ・ 13

As part of this collaboration and within the context of Saturdays.AI, our main goal is to create an introductory Brain-Computer Interfacing bootcamp to be done over the course of 14 weeks.

We ourselves will be the first students and beta users of this course, which, if successful, will be exported to the rest of the cities in the Saturdays.AI network as a new specialization path together with the usual Machine Learning, Deep Learning and Reinforcement Learning specialization paths.

The course will propose a number of readings and resources to read through the week and will introduce a related code-based challenge to solve during the hands-on team meeting on Saturdays.

Syllabus in brief

The curriculum is based primarily on the Columbia lab course, Christian Kothe’s online course, and NeuroTechX resources.

We will use OpenVive as a tool to test the experiments, MNE for the development of the experiments.

1. BCI 101: Basic Neuroscience, Basic EEG. Data collection. BCI
   Workflow.
2. Signal Processing, Hands-on with MNE (data import + preprocessing: notch filters, ICA, etc.)
3. Artifact detection & artifact repair techniques
4. Feature extractions & ML techniques. Neurofeedback & real-time BCI (OpenVibe, LSL, mne_realtime, pylsl)
5. ERP and P300.
6. Steady-State Visually Evoked Responses.
7. Detecting stress using biosignals.
8. Measuring attention using cross-brain correlations.

Experiments we would like to try in the following weeks:

• Motor-Imagery
• Emotion-Recognition
• Error-Related Potentials
• Visually Evoked Potentials
• Event Related Potentials
• Resting State
• Music and eeg
• Eye-blinks/movements

(In development)

The end goal

One of our goals is to finish the bootcamp with enough knowledge to move objects “with our minds”. Technically, we will gather EEG brainwaves using a Bitbrain 16-channel headset (as seen in the picture) and we will create machine learning models to create classifiers capable of predicting the required actions to move motor-enabled objects such as electric wheelchairs or RC cars.

Our initial experiments exploring approaches to achieve this can be found in /first-experiments

rameerez/brain-computer-interfacing

In addition, following the open source principle of Saturdays.AI, we will publish our progress with the aim that this course can be replicated anywhere in the world whether a chapter of AI Saturdays or not.

Headset

We’re using a Bitbrain EEG headset (semi-dry electrodes). All electrodes were placed following the 10–20 system.

Inspired by the work of Javier Minguez (Co-founder & CTO of Bitbrain), we have taken the first steps in this field, first detecting artifacts and then training a sorter to discern whether a subject is thinking of moving the right or the left hand. We will soon present our work.

This high-tech equipment allows us to get as close as possible to real experiments, making the learning more complete. If you do not have access to this type of headset, you can try to replicate the code and experiments with others.

Acknowledgements

We would like to thank the NeuroTechX community for helping to democratize knowledge about neurotechnology, and for serving as a resource for this program 🧠

In addition, we would like to thank AWS as a partner of Saturdays.AI for helping us bring AI around the world 🌍

Weeks: 0 ・ 1 ・ 2 ・ 3 ・ 4 ・ 5 ・ 6 ・ 7 ・ 8 ・ 9 ・ 10 ・ 11 ・ 12 ・ 13

Spread the word, follow our AI journey. You can find us on Twitter & Instagram

Email us at team@saturdays.ai