Optimization of CNN through Transferable Online Knowledge for Stress and Sentiment Classification

As we stand on the cusp of an evolution in effective and confidential smart healthcare systems, the disciplines of Psychology and Neuroscience remain a barrier. The obstacle of stress and sentiment classification is an enduring challenge to the field. Therefore, this research identifies mental health by analysing and interpreting acquired biological data. By employing convolutional neural networks in conjunction with transfer learning, the article seeks to leverage physiological signs driven by sensors for health monitoring. More precisely, we elaborate on the correlation between vital signs, arousal, and vigour data to classify a person’s sentimental state. A novel algorithmic methodology is proposed in which the source and target domains are leveraged adaptively by homogeneous and heterogeneous transfer learning. A comprehensive analysis of the outcomes from real-world acquired datasets was performed to demonstrate the proposed method’s effectiveness compared to state-of-the-art classification techniques in the field.