A Multimodal Multi-Objective Feature Selection Method for Intelligent Rating Models of Unmanned Highway Toll...

Area of Science:

• Computer Science
• Artificial Intelligence
• Machine Learning

Background:

• Intelligent classification of unmanned highway toll stations requires effective feature selection.
• Balancing feature quantity, classification accuracy, and acquisition costs is a key challenge.
• Existing methods may not adequately address the specific needs of unmanned toll station classification.

Purpose of the Study:

• To propose a novel multimodal multi-objective feature selection (MMOFS) method.
• To optimize feature selection for unmanned highway toll station classification models.
• To provide decision-makers with balanced solutions for model complexity and classification accuracy.

Main Methods:

• Utilizing a multimodal multi-objective evolutionary algorithm for feature selection.
• Employing the random forest method for the classification task.
• Evaluating the proposed MMOFS method against competing approaches using real-world data.

Main Results:

• The proposed MMOFS method demonstrated superior performance compared to two competitors.
• Performance was measured using metrics such as PSP, HV, and IGD.
• The algorithm generated multiple equivalent feature selection schemes.

Conclusions:

• The MMOFS method offers an effective approach for feature selection in unmanned highway toll station classification.
• It achieves a practical balance between model complexity and classification accuracy.
• Provides valuable guidance for the development and implementation of intelligent unmanned highway toll stations.