Demo-Training

Example training of the CNN spike ripple detector

The CNN spike ripple detector: a method to classify spectrograms from EEG data using a convolutional neural network (CNN).

---

Demo

To run a demonstration of training the CNN spike ripple detector,

• Prepare the environment
• Open and run the notebook Demo_Train_CNN.ipynb

This demonstration trains the model on data from train and valid folders, tests on data from test folder.

Data Structure

To run, you must have a data folder of the following structure:

data/

├── train/

├── Yes

├── No

├── valid/

├── Yes

├── No

├── test/

The Yes and No subfolders contain positive and negative case images on which we train the model. The test folder contains uncategorized images on which we test the model.