Wavelet ELM-AE Based Data Augmentation and Deep Learning for Efficient Emotion Recognition Using EEG Recordings

Emotion perception is critical for behavior prediction. There are many ways to capture emotional states by observing the body and copying actions. Physiological markers such as electroencephalography (EEG) have gained popularity, facial emotions may not always adequately convey true emotion. This study has two main aims. The first measure four emotion categories using deep learning architectures EEG data. second purpose increase number of samples in dataset. To this end, a novel data augmentation approach namely Extreme Learning Machine Wavelet Auto Encoder (ELM-W-AE) proposed augmentation. both simple faster than other synthetic approaches. For architectures, large datasets important performance. reason, multiplexing approaches with classical methods become popular recently. ELM-W-AE because its efficiency detail reproduction. ELM-AE structure uses wavelet activation functions Gaussian, GgW, Mexican, Meyer, Morlet, Shannon. Deep convolutional classify signals images. waves scalogram Continuous Transform (CWT). ResNet18 architecture recognizes emotions. technique GAMEMO collected during gameplay. Each these represented collection. visual set created from signal was divided into groups 70% training 30% testing. been fine-tuned augmented photos, images only. It achieved 99.6% classification accuracy tests. method compared on same dataset, an approximately 22% performance improvement achieved.