Emotion Recognition of GSR Based on an Improved Quantum Neural Network

• Published in: 2016 8th International Conference on Intelligent Human-Machine Systems and Cybernetics (IHMSC) Vol. 1; pp. 488 - 492
• Main Authors: Qifei Zhang, Xiangwei Lai, Guangyuan Liu
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.08.2016
• Subjects: Biological neural networks; Convergence; Emotion recognition; Feature extraction; GSR signal; Particle swarm optimization; Quantum neural network; Training
• DOI: 10.1109/IHMSC.2016.66

Five kinds of emotions including happy, fear, grief, anger and calm were induced by experimental environment, the galvanic skin response(GSR) signals of five kinds of emotions were effectively collected as samples from 35 subjects, and a small fragment of GSR signal was intercepted from each sample data to make up sample database which contained 175 samples, then the processed data was used to emotion classification after preprocessing GSR signals and extracting 30 features of the GSR signals. Since each feature is not fully able to reflect changes in five kinds of emotion, and there are overlaps in the optimal feature sets of GSR signal for different emotions, the paper constructed a quantum neural network recognition model for recognizing different emotions. Furthermore, the training algorithm of the traditional quantum neural network is easy to fall into local optimum and has poor convergence performance. This paper proposed an improved quantum neural network based on particle swarm optimization algorithm. Experimental results showed that the performance of quantum neural network based on improved particle swarm optimization was better than that of the quantum neural network based on the conventional gradient descent.