The effect of activation functions in MLP performance based on different classification cases / Iza Sazanita Isa, Siti Noraini Sulaiman, Azizah Ahmad, Normasni Ad. Fauzi and Nurul Huda Ishak

Multilayer perceptron network (MLP) has been recognized as a powerful tool for many applications including classification. Selection of the activation functions in the multilayer perceptron (MLP) network plays an essential role on the network performance. This paper presents comparison study of different MLP activation function; for three different classification cases which are breast cancer, thyroid disease and weather classification. The activation functions under investigation are sigmoid and hyperbolic tangent. In this study, the MLP network was trained and tested to investigate the ability of network to classify the breast cancer between benign and malignant cell, thyroid disease are classified into normal, hyper or hypo thyroid while the weather conditions are classified into four types; rain, cloudy, dry day and storm. Levenberg-Marquardt algorithm is used to train the MLP network since it is the fastest training and ensure the best converges towards a minimum error.