Cancers classification based on deep neural networks and emotional learning approach

Area of Science:

• Computational biology
• Machine learning in oncology

Background:

• Cancer classification traditionally relies on expert opinion, but evolving complexities necessitate intelligent algorithms.
• The need for robust and accurate automated cancer classification methods is growing.

Purpose of the Study:

• To propose a novel algorithm for classifying colon, ALL-AML, and leukaemia cancer datasets.
• To integrate deep neural networks and emotional learning for improved cancer classification accuracy.

Main Methods:

• Feature reduction using Principal Component Analysis (PCA).
• Feature extraction via deep neural networks.
• Classification using Multi-Layer Perceptron, Support Vector Machine (SVM), Decision Tree, and Gaussian Mixture Model (GMM).
• Enhancing model robustness against uncertainty using a stacked denoising autoencoder with Gaussian noise injection.

Main Results:

• Emotional learning generally improved classification accuracy across datasets.
• SVM achieved the highest accuracies: 91.66% (colon), 92.27% (ALL-AML), and 96.56% (leukaemia).
• GMM demonstrated the lowest accuracy, with the best performance at 60%.

Conclusions:

• The proposed deep learning algorithm, enhanced by emotional learning, shows promise for accurate cancer classification.
• SVM emerges as a highly effective classifier within this framework.
• The method's robustness against uncertainties is a key feature for real-world applications in systems biology.