Logo System Theory Group

Signal & System Theory Group

Professor Peter Schreier
Universität Paderborn, Germany

Emotion recognition based on EEG and signal processing

Background

Figure 1: Emotiv EEG headset

Electroencephalograms (EEG) devices have been widely used in the past for effective clinical and healthcare services. However, with new and portable equipments such as Emotiv headset (Fig. 1), EEG devices are being applied in our day to day lives for emotion detection, getting consumer insights and brain computer interfacing with widespread applications such as tourism, virtual reality, urban planning and business decision making (see emotiv.com/solutions for more details). Moreover, with the advent of improved signal processing and machine learning tools, we are better equipped for processing, classifying and learning the data generated from these devices.

Task

Figure 2: Data recorded from Emotiv software

We propose to design an emotion recognition system using brain signals only. While emotion recognition systems based on speech and facial expressions have been developed in the past, they can be easily fooled by the user by concealing their true emotions. On the other hand, the brain activity due to emotional changes cannot be voluntarily concealed.

You can collect your own brain data after applying an audio or video stimulus to arouse different emotions. This data can be analyzed and processed to extract meaningful features, for example, by applying Fourier transformation to acquire brain activity at different frequency bands. The extracted features can then be fed to different machine learning tools such as support vector machines (SVM), neural networks (NN) to train and classify the data into various emotional states.

Prerequisites
Knowledge of digital signal processing and basic knowledge about statistical signal processing and machine learning. Programming skills in Matlab/Python.