An Efficient Approach to Breast Cancer Prediction Based on Neural Network, Adaboost and Gaussian Process

摘要

In this paper, we present a hybrid of Incremental Learning radial basis function Neural Network, Gaussian Process classifier and AdaBoost for building a breast cancer survivability prediction model. Diagnosis of breast cancer is a difficult to accomplish due to its noisy data and relatively small database. We carried out our experiment on breast cancer Wisconsin data base from the UCI repository which has 16 missing values. We applied Gaussian process regression for predicting of missing value attributes. Then, we combined RBF and AdaBoost, and Gaussian Process classifier algorithms to develop a novel classifier. The capability of this method is evaluated using a 10-fold cross validation. Experimental results on this data set reveals that the suggested method provides higher prediction accuracy than conventional classifiers.