A Multi-Column CNN Model for Emotion Recognition from EEG Signals

Abstract

We present a multi-column CNN-based model for emotion recognition from EEG signals. Recently, a deep neural network is widely employed for extracting features and recognizing emotions from various biosignals including EEG signals. A decision from a single CNN-based emotion recognizing module shows improved accuracy than the conventional handcrafted feature-based modules. To further improve the accuracy of the CNN-based modules, we devise a multi-column structured model, whose decision is produced by a weighted sum of the decisions from individual recognizing modules. We apply the model to EEG signals from DEAP dataset for comparison and demonstrate the improved accuracy of our model.

A Hybrid Feature Pool-Based Emotional Stress State Detection Algorithm Using EEG Signals

“…After considering the 32 EEG channels, the valence and arousal levels are analyzed from the online self-assessment of each participant for each experiment ID [18,19]. Every experiment ID has a predefined online rating, by which all the experiment IDs can be categorized either the genre of stress state or the genre of calmness.…”

Distinguishing Emotional Responses to Photographs and Artwork Using a Deep Learning-Based Approach

“…Yang et al [16] presented a multi-channel structured-CNN model to recognise emotions from unstationary EEG signals. Their model is composed of several independent recognition modules, which are designed based on the DenseNet model [17].…”

“…We employ a multi-column structured model, which shows state-of-the-art accuracy in recognizing emotion from EEG signals [16]. This model is composed of several recognizing modules that process the EEG signal independently.…”

“…The whole structure of our model is illustrated in Figure 6b. According to [16], the best accuracy is achieved for a model of five recognizing modules. Therefore, we built our model with five recognizing modules.…”

“…k is the number of recognition modules in our multi-column emotion recognition model. Yang et al [16] tested four values for k: 1, 3, 5 and 7. Among the four values of k, k = 5 shows the highest accuracy for valence and the second highest accuracy for arousal.…”