Artificial Intelligence approaches have been one of the most widely used tools for prediction and classification in many fields for cases where we use deductive reasoning. Specifically, in the medical field, artificial intelligence and neural networks have been used for diagnosing patients by using pattern recognition, prediction, and classification methods. Among many medical applications, the classification of breast cancer has been one of the important topics for researchers. Finding a classification approach with high reliability and a low error within the testing data-set has been one of the greatest challenges among researchers and scientists. One of the disadvantages of a neural network classifier is that it will be stuck at the local minimal unable to reach an accurate result due to possible outliers within data-set. This may happen if the data-set in use contains outliers which affect the proper convergence of the neural network. To overcome this problem, the detection and removal of the outliers within the data-set is proposed. The objective of this study is to use the neural network classification to diagnose breast cancer for the data-set after eliminating outliers. This can be used for normal diagnosis of cancer without considering special cases. Our approach uses the UCI data-set by implementing the neural network and detecting and removing the outliers. We run the neural network on the remaining data-set to test the performance of the classifier.
