Voting based double-weighted deterministic extreme learning machine model and its application.

Area of Science:

• Machine Learning
• Artificial Intelligence
• Computational Biology

Background:

• Traditional Extreme Learning Machine (ELM) models face limitations due to input layer weights and hidden layer bias, leading to large neuron counts and unstable performance.
• These limitations hinder the effectiveness of ELM in complex classification tasks.

Purpose of the Study:

• To introduce an improved intelligent learning model, the voting-based Double Pseudo-inverse Extreme Learning Machine (V-DPELM).
• To address the instability and performance issues associated with traditional ELM methods.
• To enhance classification accuracy in real-world datasets, particularly for medical applications like breast tumor recognition.

Main Methods:

• Development of the voting-based Double Pseudo-inverse Extreme Learning Machine (V-DPELM) model.
• Direct determination of weight structure and implementation of a voting mechanism strategy.
• Extensive simulations and comparative analysis against traditional V-ELM methods on diverse classification datasets.

Main Results:

• The V-DPELM model demonstrates significantly improved classification accuracy compared to traditional V-ELM methods.
• The proposed model effectively alleviates the limitations of traditional methods, showing more stable performance.
• Superior classification accuracy was achieved when V-DPELM was applied to machine recognition of breast tumors.

Conclusions:

• The V-DPELM model offers a robust and accurate solution for classification tasks.
• Its enhanced performance makes it a valuable tool for machine-assisted diagnosis, especially in identifying breast tumors.
• The V-DPELM model represents a significant advancement in intelligent learning for classification problems.