In “Brain Waves,” a neuroscience high school program that was developed at New York University, students use EEG to learn about their own brains, and about how neuroscience works. Study of the brain in natural situations can benefit our understanding of social interactions. When we found out that there was a way we can build something which can help us know how our brain processes different emotions, and to be able to incorporate it in our own systems, nothing was keeping us away from it.

What it does

Psyce predicts emotions (Positive/Neutral/Negative) depicted by the input EEG brain waveform. The input is a CSV file of 2549 columns representing the time-series format of the actual brainwave and classifies the emotion depicted by it into three categories If the emotion detected is negative, you get a small surprise to make you feel better.

How we built it

We have created a deep learning, RNN model that examines the EEG report and classifies the emotion of a person as negative, positive or neutral. We have used ‘EEG Brainwave Dataset: Feeling Emotions’ dataset from Kaggle, which has 2132 rows x 2549 columns. Each row of the dataset represents the time-series format of a brain wave. The RNN model consists of an Input Layer, a GRU layer and an output layer Dense with softmax activation function. The above model performed at an accuracy of 96%. We have further created a user friendly interface using the Flask API and have also incorporated Notivize to alert you through email notifications.

Challenges we ran into

Unavailability of portable EEG headsets, which could've proven to be an asset for real-time application for Psyce. Training the model to yield maximum accuracy. Using Flask API to provide an interface for user interaction.

Accomplishments that we're proud of

We were able to incorporate Notivize in our project. We also got to add so much to our knowledge of deep learning and artificial intelligence as it was our first time working with RNN.

What we learned

About brain waves and how our brain functions in its natural environment. About RNN To deploy an end-to-end Machine Learning model.

What's next for Psyce

We plan to take Psyce one step ahead by incorporating a portable EEG headset which can act as a direct input to our model.

Share this project: