Active Learning in Emotive Intelligent Spaces

2019-2020

The Machine Learning model that was constructed using the wristband data is used to help gather and label EEG data. The EEG is not portable and requires connection to a PC so the data collection is in a laboratory setting.  Participants wore both the wristband and EEG helmet. They were asked to watch any video of their choosing to evoke a particular emotional response.

An Active Learning method was used to record and label the EEG data. A Python program runs the learned wristband model and makes predictions of the emotion the participant is feeling at random intervals. The participant is prompted to either confirm that they are feeling the predicted emotion or to enter the correct emotion if the prediction is incorrect. The sensor measurements are in microvolts and are read sequentially by a Python program. The emotion label is concatenated to the sensor data and the labelled EEG data is then stored in a comma separated text file to be used to train a neural network.  

Year: 2019-2020  

Team: Mona Ghandi, Sal Begavav, Alex Trevithick, Marcus Blaisdell