Roadmap

Goal: develop an open platform with educational resources for algorithmic decision-making in neuroscience

Here is a list of milestones for this project.

Milestone: List of tutorials for EEG data

How to apply common algorithms on electroencephalography (EEG) data?

• Loading and visualization of epoch data
• Pre-processing: What are common artifacts? How to reject them? Filtering.
• Statistics: Contrasts between two conditions at single-electrodes. Permutation statistics.
• Machine learning: What features can we have? How to implement cross-validation?
• Machine learning: Common algorithms: LDA / Logistic Regression
• Machine learning: How to visualize features and classification accuracy
• Expand to a different dataset / application

Milestone: Tutorials for signal processing

e.g. filters, time-frequency transforms

• Filters...
• Time-frequency

Milestone: Bind everything together

e.g. through mybinder or Jupyter book

• Test that MyBinder works
• Implement everything in Jupyter book
• Proof read documentation

Milestone: Community building

e.g. gather contributions and expand to other techniques

• Get feedback from neuroscientists
• Seek additional contributors
• Get feedback from students

1. Are the tutorials clear? (if they are not, please give the names of tutorials while commenting)

2. Have you found anything that is incorrect or missing in the contents of the tutorials?

3. Which topics would you like to see in the upcoming tutorials?

Note: Please leave a comment below and don't forget to indicate question numbers that you are answering and tutorial names. (names are only required for questions 1 & 2)

Notebook: ApplyingMachineLearningMethods_1:

At the first step, we will simply load epoched EEG data, which here have been converted to an MNE format. To obtain more information about this format, please refer to the tutorial named 'study1'. <-- we need to add a link

List of tutorials in machine learning

This is just a preliminary list, we can always elaborate with additional resources.

• Different representations of EEG data that can be used for classification
• Contrast of different feature selection techniques: PCA, ICA, time-frequency, etc
• Comparison for different machine learning algorithms on a sample dataset of 1 participant
• Contrasting single-participant and group-level classifiers
• Visualizing features of an accurate classifier.

Trying the tutorials with different datasets would help improving the existing tutorials by giving hints on special cases etc. Also, for machine learning tutorials, different objectives can be examined on various datasets and this would help us to provide more generalised series of tutorials.

Following is a tentative roadmap for DL for EEG Tutorials.

• Introduction to DL (DNN, CNN & RNN) with EEG data and famous DL frameworks (tensorflow & pytorch)
• Selecting a network
• Import and Preprocess Data for DNN
• Configuring, Training and Testing a DNN Model
• Import and Preprocess Data for CNN
• Configuring, Training and Testing a CNN Model
• Import and Preprocess Data for LSTM
• Configuring, Training and Testing an LSTM Model
• Visualising and Making sense of predictions

This is a tentative framework on which @binarycache and I will start working.

Changes and additions to this are welcome.

This will serve as a preliminary table of contents

Hi,
Great to see this work, I was looking to see if I could join the Slack Channel, but it looks like the link only allows sign in and not for new registration into the workspace?
Would be great if you could update it or tell me if I'm going about it wrong.

Best,
Divyesh Narayanan

Include project description and how contributors can reach us

Create a guideline for installing the packages and libraries that are used throughout the tutorials