[Study on Stellar Spectral Outliers Mining Based on Fuzzy Large Margin and Minimum Ball Classification Model].

Area of Science:

• Astronomy and Astrophysics
• Data Science
• Machine Learning

Background:

• Discovering unique celestial bodies is a key objective in universe exploration.
• Traditional classification methods struggle with outlier spectral data, hindering the identification of special celestial bodies.
• Existing approaches lack sensitivity to outlier data, impacting overall classification efficiency.

Purpose of the Study:

• To develop a novel classification model for enhanced exploration of special celestial bodies.
• To address the limitations of traditional methods in handling outlier spectral data.
• To improve the efficiency and accuracy of celestial body classification.

Main Methods:

• Proposed the Fuzzy Large Margin and Minimum Ball Classification Model (FLM-MBC).
• Utilized fuzzy techniques to mitigate noise influence on classification.
• Trained the model using both general and outlier spectral data to construct a minimum ball model.

Main Results:

• The FLM-MBC model demonstrated superior performance compared to traditional methods.
• Comparative experiments confirmed the effectiveness of FLM-MBC on SDSS spectral datasets.
• The proposed model shows significant improvements in classifying outlier spectral data.

Conclusions:

• FLM-MBC is an effective tool for identifying special celestial bodies through spectral outlier analysis.
• The model offers a robust solution for noise reduction in astronomical data classification.
• This approach enhances the capabilities of universe exploration by improving celestial body discovery.