An Eye Movement Classification Method Based on Cascade Forest.

Area of Science:

• Human-Computer Interaction
• Neuroscience
• Machine Learning

Background:

• Eye tracking technology is vital for scientific research and applications.
• Classifying raw eye movement data into events is crucial but challenging.
• Existing methods struggle with participant variability, class imbalance, and data scarcity.

Purpose of the Study:

• To introduce a novel eye movement classification method, EMCCF, based on cascade forest.
• To enhance the accuracy and efficiency of eye movement classification.
• To advance the application of ensemble learning in eye movement analysis.

Main Methods:

• A feature extraction module using a multi-scale time window.
• A classification module employing a layered ensemble architecture with cascade forest.
• Integration of ensemble learning principles for robust classification.

Main Results:

• EMCCF demonstrated enhanced accuracy and efficiency in eye movement classification.
• The method showed robust performance across diverse datasets and participants.
• EMCCF outperformed existing deep learning-based classification models in key metrics.

Conclusions:

• EMCCF offers a significant advancement in eye movement classification.
• The novel cascade forest approach effectively addresses data scarcity and class imbalance.
• This method provides a more adaptable and efficient solution for analyzing eye movement data.