Multi-Layer Perceptron

The application of machine learning technologies for cancer detection purposes are rising due to their ever-increasing accuracy. Melanoma is one of the most common types of skin cancer. Detection of melanoma in the early stages can significantly prevent illness and fetal death. The application of innovative machine learning technology is highly relevant and valuable due to medical practitioners' difficulty in early-stage diagnoses. This paper provides an open-source tutorial on the performance of an algorithm that helps to diagnose melanoma by extracting features from dermatoscopic images and their classification. First, we used a Dull-Razor preprocessing method to remove extra details such as hair. Next, histogram adjustments and lighting thresholds were used to increase the contrast and select lesion boundaries. After using a threshold, a binary-classified version of image was obtained, and the boundary of the lesion was determined. As a result, the features from skin tissue were extracted. Finally, a comparative study was conducted between three methods which are Artificial Neural Network (ANN), Support Vector Machine (SVM) and K-Nearest Neighbor (KNN). The results show that ANN could achieve better accuracy (83.5%). In order to mitigate the biases in existing studies, the source code of this research is available at hadi-naghavipour.com/ml to serve aspiring researchers for improvement, correction and learning and provide a guideline for technology manager practitioners.

No unique method has been so far specified for determining the number of neurons in hidden layers of Multi-Layer Perceptron (MLP) neural networks used for prediction. The present research is intended to optimize the number of neurons using two meta-heuristic procedures namely genetic and hill climbing algorithms. The data used in the present research for prediction are consumption data of water subscribers in Fasa City of Fars Province (Iran) between the years 2010 to 2013. Ultimately, using the respective data set, the data of the subsequent year 2014 can be predicted. In the present research it was observed that the mean square errors of per data (MSEPD) for the abovementioned algorithms are less than 0.2, indicating a high performance in the neural networks’ prediction. Correlation coefficients using genetic and hill climbing algorithms were respectively equal to 0.891 and 0.759. Thus, GA was able to leave a better effect on optimization of neural network.