A Hybrid Model Based on Deep XGBOOST for Creating a Stable Network and Classifying Cancer Tumor Images Using Supervised Blended Learning (Cono-XGBoost)

Blended learning is expanding and is being employed to solve various problems. Network stabilization and incremental network learning (enhancing weak learners) are some benefits that can be pursued in this type of learning. Moreover, the application of blended learning in medical science has also increased with the advancement of machine learning, which is used in various aspects of this field, such as diagnosis, treatment, and prevention. Machine learning in medical image processing has undergone significant development over the past few years. The classification of medical images, including the detection of cancer tumors, fractures, masses, etc., has been among these research endeavors. This study aimed to utilize the development of the XGBOOST method to propose a technique for reducing the error rate in the classification of cancer tumor images. In addition to error reduction, another goal pursued in this model is network stabilization to enhance the effectiveness of identifying various images. The research, evaluated on different datasets, shows a 2% reduction in error compared to previous methods. This technique can potentially be used in the future to classify other topics.