ECG-Based Automated Emotion Recognition Using Temporal Convolution Neural Networks

This study introduces a novel application of temporal convolutional neural networks (TCNNs) for automated emotion recognition (AER) using electrocardiogram (ECG) signals. By leveraging advanced deep learning (DL) techniques, our approach achieves the impressive classification accuracies of 98.68% for arousal and 97.30% for valence across two publicly available datasets. This methodology effectively preserves the temporal integrity of ECG signals, offering a robust framework for real-time emotion detection. Extensive preprocessing ensures high-quality input data, while cross validation confirms model generalizability. Our results demonstrate the potential of TCNN in enhancing human-computer interactions and healthcare monitoring systems through improved emotion recognition, paving the way for future applications in affective computing and wearable sensor technology.