This talk will be about How to process EEG signals to make a comprehensive Brain-Computer Interface (hereafter referred as BCI) system.

Electroencephalogram (EEG) is perhaps one of the simplest and easy ways to understand brain activities. EEG records the electrical signals produced by neurons amplifies them and show them as waveforms. If we can understand this waveform, then we can identify what our brain trying to do. Example: Consider an operation on the music system, say increasing and decreasing volume. If we can learn the waveform for volume increase and volume decrease, we can control volume without actually touching any device. This becomes the fundamental idea behind BCI

Feature extraction is a key aspect of any machine learning use case, in case of signal processing it becomes even more complicated as we don’t have any means to visualize it, thus comes several mathematical concepts and theorems which help us in analyzing it.

As part of this talk, I will be covering several mathematics concepts like Fourier transformation, wavelets, Fourier convolutions, etc.… which help in understanding generated signals

I will be talking about available python libraries which we can use in our applications. I will be showing code snippets and plot graphs for better understanding.

I will be touching upon some of the aspects on How to build a Machine learning model for the prediction. A model that is aligned to BCI (using the neural network) and How to evaluate the model

5 favorite thumb_down thumb_up 6 comments visibility_off  Remove from Watchlist visibility  Add to Watchlist

Outline/Structure of the Talk

In a high level,

There will be talk, wherein I will cover how to extract features using Fourier transformation wavelets. I am planning to show code snippets and graphs to aid better visualization.

I will talk about what technique to use to extract data, when to use that technique, what is the right way of using it, how to decide which technique to use. (This is an extension of previous point)

I'll be talking about how to build a NN for BCI and advantages of it over other ML models


And demo if possible or if folks are interested

Learning Outcome

A person will have right direction when it comes to EEG processing. It's like a compass pointing in the right direction

Target Audience

People who want to work on signal processing but have very less knowledge of how to extract features, How to decide which feature or model to choose etc..People who want to work on signal processing but have very less knowledge of how to extract features,

Prerequisites for Attendees

They should have basic machine learning knowledge. Since I will be covering basics of signal processing, Domain knowledge is not mandatory

schedule Submitted 9 months ago

Public Feedback

comment Suggest improvements to the Speaker
  • Vishal Gokhale
    By Vishal Gokhale  ~  9 months ago
    reply Reply

    Thanks for the proposal, Raghavendra ! :-)

    This is an interesting topic. 
    Would it be possible to present it in the following form:

    1. A relatable example problem
    2. Solution approaches
    3. Chosen solution approach
    4. Challenges, if any and solutions using techniques like Fourier Transform, Wavelets etc.
    5. Code examples / demos
    6. Tying it all back together to show the complete picture of the example
    7. 'Read' the abstract idea hidden in the example/solution approach and show how it can be generalized for more applications. 
    • Ashutosh Tiwari
      By Ashutosh Tiwari  ~  8 months ago
      reply Reply

      Great article dude .i really liked it