Stress emotion classification using optimized convolutional neural network for online transfer learning dataset.

Area of Science:

• Computer Science
• Artificial Intelligence
• Machine Learning

Background:

• Deep learning with transfer learning (TL) simplifies stress and emotion classification.
• Optimized Convolutional Neural Network with TL (OCNNTL) learns high-level features but struggles with real-time concept drift and data processing.

Purpose of the Study:

• To propose a novel online OCNNTL (O2CNNTL) model to address limitations of existing methods.
• To effectively handle abrupt concept drift in real-time stress and emotion classification.

Main Methods:

• The O2CNNTL model leverages prior knowledge from stress and emotion domains for initial classification.
• It incorporates an OCNN classifier to manage concept-drifting data streams in real-time.
• A regularization learning technique is applied across varied feature spaces to boost learning efficiency.

Main Results:

• The O2CNNTL model demonstrates superior performance in handling concept drift compared to traditional OCNNTL.
• It achieves higher classification efficiency, outperforming state-of-the-art models.
• The model reduces time complexity associated with data gathering and transformation.

Conclusions:

• The proposed O2CNNTL model offers an effective solution for real-time stress and emotion classification.
• It successfully overcomes the limitations of abrupt concept drift and data processing challenges.
• O2CNNTL presents a more efficient and robust approach for dynamic classification tasks.