3.3.5. Support Vector Machines (SVM)

Support Vector Machines (SVM) is a model that works by finding a “hyperplane”, as of an invisible line or boundary, that separates different groups in the data. SVMs are able to work in higher dimensions through the use of “kernel transformation” techniques such as linear, polynomial, and radial basis function (RBF) kernels. The idea is same as lifting data to a new view where patterns become more apparent and easier to be divided. SVMs are effective for emotion recognition and capturing subtle differences in EEG and ECG signals, though they may struggle with large or noisy datasets and require careful preprocessing for better results.

Zhang et al. implemented SVM with multi-view learning to recognize surprise and calm emotional states in piplots by fusing EEG and ECG data, achieving effective classification under simulated flight conditions . Xue et al. integrated olfactory stimuli with EEG recordings to enhance emotion detection, with the SVM classifier achieving a notable 99.03% accuracy in classifying emotions under video-odor stimuli . Hasnul et al. compared SVM-based performance on the DREAMER and AuBT datasets, finding SVM responsive to different feature extraction methods for arousal and valence classification . Lastly, Riaz et al. explored emotion recognition in response to high dynamic range (HDR) video stimuli, achieving classification accuracies of 80.55% for arousal and 70.37% for valence using SVM on EEG data .