A Novel Extreme Learning Machine Classification Model for e-Nose Application Based on the Multiple Kernel Approach.

Area of Science:

• Machine Learning
• Computational Intelligence
• Chemometrics

Background:

• Electronic noses (e-noses) are crucial for gas classification.
• Existing multiple kernel extreme learning machine (MK-ELM) algorithms have limitations in optimizing kernel combinations.
• Accurate and efficient classification models are needed for e-nose data.

Purpose of the Study:

• To propose a novel classification model, the quantum-behaved particle swarm optimization (QPSO)-based weighted multiple kernel extreme learning machine (QWMK-ELM).
• To optimize combination coefficients, base kernel parameters, and regularization parameters simultaneously using QPSO.
• To validate the QWMK-ELM model's performance on electronic nose datasets.

Main Methods:

• The QWMK-ELM model treats base kernel combination coefficients as external parameters of single-hidden layer feedforward neural networks (SLFNs).
• Quantum-behaved particle swarm optimization (QPSO) is employed to simultaneously optimize kernel combination coefficients, base kernel parameters, and the regularization parameter.
• Composite kernels are constructed using four common single kernel functions: Gaussian, polynomial, sigmoid, and wavelet.

Main Results:

• The QWMK-ELM model demonstrated superior performance compared to existing methods like ELM, KELM, KNN, SVM, MLP, RBFNN, and PNN.
• Experimental validation on two e-nose datasets confirmed the model's enhanced precision.
• The proposed QWMK-ELM also showed significant improvements in classification efficiency.

Conclusions:

• The QWMK-ELM model offers a significant advancement in gas classification using electronic nose technology.
• Simultaneous optimization via QPSO effectively enhances the performance of weighted multiple kernel extreme learning machines.
• The QWMK-ELM provides a more precise and efficient solution for complex e-nose data analysis.