Enhancing Electronic Nose Performance by Feature Selection Using an Improved Grey Wolf Optimization Based Algorithm.

Area of Science:

• Artificial intelligence
• Chemical sensing technology
• Data science

Background:

• Electronic noses (e-noses) are artificial olfactory systems for detecting volatile organic compounds (VOCs).
• High-dimensional data from e-noses presents challenges for pattern recognition and gas classification.
• Effective feature selection is crucial for developing accurate e-nose models.

Purpose of the Study:

• To propose an improved Grey Wolf Optimizer (GWO) algorithm for feature selection in e-nose data.
• To enhance the accuracy and efficiency of gas recognition models.
• To address the challenge of high-dimensional data in e-nose applications.

Main Methods:

• Developed an improved GWO algorithm incorporating novel binary transform approaches for feature subset searching.
• Integrated an adaptive restart mechanism to boost algorithm search capability and stability.
• Applied and evaluated the algorithm on three distinct e-nose datasets using three classifiers and multiple assessment metrics.

Main Results:

• The proposed GWO-based algorithm effectively identified feature subsets that significantly improve gas recognition.
• Experimental comparisons demonstrated superior performance over five other feature selection algorithms.
• The method successfully enhanced the overall performance of the electronic nose system.

Conclusions:

• The improved GWO algorithm is a robust and effective tool for feature selection in e-nose applications.
• This approach optimizes data dimensionality, leading to more accurate volatile organic compound detection and classification.
• The study highlights the potential of advanced optimization algorithms in enhancing artificial olfactory systems.