Optimal Scale Combination Selection Integrating Three-Way Decision With Hasse Diagram.

Area of Science:

• Machine Learning
• Data Mining
• Artificial Intelligence

Background:

• Multi-scale decision systems (MDS) are crucial for analyzing hierarchical data in machine learning.
• Optimal scale combination (OSC) selection and attribute reduction are key challenges in MDS knowledge discovery.
• Existing methods for OSC selection often suffer from combinatorial explosion and excessive time consumption.

Purpose of the Study:

• To develop an efficient method for searching all optimal scale combinations (OSCs) in multi-scale decision systems (MDS).
• To reduce the computational complexity and time required for knowledge discovery in MDS.
• To integrate three-way decision models with Hasse diagrams for optimized scale space searching.

Main Methods:

• A novel scale combination approach is proposed for simultaneous scale selection and attribute reduction.
• An extended stepwise optimal scale selection (ESOSS) method is introduced for rapid local OSC identification.
• A sequential three-way decision model of the scale space is established to partition the search space.
• The Hasse diagram properties are utilized to calculate maximal elements, reducing space complexity.

Main Results:

• The proposed method effectively reduces the search space for OSCs by dividing it into positive, negative, and boundary regions.
• Local OSCs found in the boundary region are proven to be global OSCs.
• The algorithm efficiently identifies all OSCs by iteratively searching boundary regions.
• Experimental results show a significant reduction in computational time compared to existing approaches.

Conclusions:

• The developed algorithm provides an efficient solution for selecting all optimal scale combinations in MDS.
• The integration of three-way decision and Hasse diagrams offers a novel approach to optimize search spaces.
• The method significantly enhances the efficiency of knowledge discovery in hierarchical data analysis.