Abstract:

Breast cancer is one of the most dangerous cancers that causes many deaths every year. Early diagnosis plays a constructive role in cancer treatment. Therefore, researchers are conducting experimental research using methods of classification and accumulation of data about patients and healthy people. In addition to the advanced technology and machine learning-backed diagnostic studies, a remarkable increase in the use of deep learning methods is seen. In this study, a new model was designed to be used in breast cancer classification using accumulated identifiable coders (SAE), a deep learning method. The most commonly used machine learning methods to make performance comparison with the SAE designed support vector machines, k-the closest neighbourhood, naive bayes and decision trees methods have also been used in this study. In addition to the accuracy ratio metric, the time spent in the training and test stages (time complexity) is calculated in experimental studies. In experimental studies, the impact on the success of classification, by applying the normalization process from the data pre-processing steps, has been studied. According to the experimental results, the accuracy ratio is based on the criteria with the 79.31% accuracy ratio and the most successful result is obtained with the SAE supported data pre-processing. According to time complexity meters, the KNN algorithm is the fastest algorithm in the training process, while the SAE algorithm is identified as the fastest algorithm in the test process.

Keywords: