Weighted ensemble model for image classification.

Area of Science:

• Artificial Intelligence
• Medical Imaging
• Machine Learning

Background:

• Deep Convolutional Neural Network (DCNN) models are widely used in medical science for image classification.
• Model accuracy and reliability are crucial for practical applications in machine learning and deep learning.

Purpose of the Study:

• To develop a heterogeneous ensemble approach using DCNN models for enhanced medical image classification.
• To improve the accuracy and reliability of DCNN models by weighting their contributions based on individual performance.

Main Methods:

• A novel DCNN-based heterogeneous ensemble method was proposed.
• Multiple DCNN models were trained on a single dataset.
• Each model's contribution to the final output was weighted according to its accuracy.

Main Results:

• The ensemble approach demonstrated a significant increase in 3-class accuracy on two COVID-19 X-ray image datasets.
• Weighted contributions ensured that more accurate models had a greater impact on the final classification.
• The proposed method outperformed existing models in literature for the tested datasets.

Conclusions:

• The DCNN-based heterogeneous ensemble method offers a robust strategy for improving medical image classification accuracy.
• Weighted ensemble models provide a reliable approach for leveraging multiple DCNNs effectively.
• This technique shows promise for applications requiring high accuracy in medical diagnostics.