Parametric optimization and comparative study of machine learning and deep learning algorithms for breast cancer...

Area of Science:

• Oncology
• Computer Science
• Data Mining

Background:

• Breast cancer is a leading cause of mortality in women.
• Manual diagnosis is time-consuming and limited.
• There is a need for automated early-stage cancer detection systems.

Purpose of the Study:

• Investigate machine learning model performance for breast cancer diagnosis.
• Compare artificial neural network (ANN) methodology with conventional techniques.
• Evaluate models on the Wisconsin Breast Cancer (original) dataset.

Main Methods:

• Applied various machine learning models including SVM, Decision Tree, CART, ANN, and ELM ANN.
• Utilized hyperparameter tuning to optimize model performance.
• Employed boosting algorithms such as XGBoost, Adaboost, and Gradient Boost.

Main Results:

• Several classifiers achieved high accuracy, precision, and F1 scores for benign and malignant tumors.
• Models with hyperparameter adjustment outperformed those without.
• Boosting methods consistently performed well in distinguishing tumor types.

Conclusions:

• Machine learning, particularly with hyperparameter tuning and boosting algorithms, is effective for breast cancer diagnosis.
• These techniques can address data complexity and nonlinearity.
• The study provides a summary of current research on breast cancer diagnosis using the Wisconsin dataset.