MLP network for lung cancer presence prediction based on microarray data

The appearance of the Microarray technology has attracted the scientific community and industry; with its ability of measuring simultaneously the activity and interactions of thousands of genes. This advanced technology was applied for enormous issues such as drug discovery, gene discovery, diagnosis and prognosis of disease and toxicological research. Despite the fact that Microarray applications have known birth in many biological studies, the handling and analysis of the data obtained are not trivial tasks. For these reasons, it has been focused on the present paper on the PCA classification technique and Neural Network for Microarray data; in the object of reducing the large data and producing informative results. The methodology proposes an approach based on MLP neural network to resolve the problem of lung cancer classification based on Microarray data. The approach consists on data reduction by using the PCA Technique, followed by a classification based on MLP network, feed-forward neural network known by its stable learning. The effectiveness of the implemented method was evaluated by measuring the correct classification rate performed on lung cancer gene expression dataset and compared to results obtained by other methods that use the same data.