Combination of signal segmentation approaches using fuzzy decision making

Area of Science:

• Signal Processing
• Machine Learning
• Biomedical Engineering

Background:

• Signal segmentation is crucial for effective signal analysis, impacting feature extraction and classification.
• Existing signal segmentation methods exhibit variable performance with changing signal characteristics.
• The need for robust and adaptable signal segmentation techniques is evident.

Purpose of the Study:

• To propose two novel algorithms for signal segmentation.
• To combine existing signal segmentation approaches using probability and fuzzy concepts.
• To enhance the efficiency of subsequent signal analysis steps like feature extraction and classification.

Main Methods:

• Development of two new algorithms integrating probability and fuzzy logic.
• Combination of several well-established signal segmentation techniques.
• Validation using synthetic and real-world electroencephalogram (EEG) signals.

Main Results:

• The proposed algorithms demonstrate enhanced performance in signal segmentation.
• The novel approaches show efficiency with both synthetic and real EEG data.
• Improved accuracy in segmentation leads to better downstream analysis.

Conclusions:

• The developed probability and fuzzy-based algorithms offer a robust solution for signal segmentation.
• These novel methods improve the overall efficiency of signal analysis pipelines.
• The findings are particularly relevant for electroencephalogram signal processing.