Artificial neural networks for artificial intelligence

Artificial neural networks now have a long history as major techniques in computational intelligence with a wide range of applications for learning from data. There are many methods developed and applied so far, from multiplayer perceptrons (MLP) to the recent ones being deep neural networks and deep learning machines based on spiking neural networks. The paper addresses a main question for researchers and practitioners: Having data and a problem in hand, which method would be most suitable to create a model from the data and to efficiently solve the problem? In order to answer this question, the paper reviews the main features of the most popular neural network methods and then lists examples of applications already published and referenced. The methods include: simple MLP; hybrid systems; neuro-fuzzy systems; deep neural networks; spiking neural networks; quantum inspired evolutionary computation methods for network parameter optimization; deep learning neural networks and brain-like deep learning machines. The paper covers both methods and their numerous applications for data modelling, predictive systems, data mining, pattern recognition, across application areas of engineering, health, robotics, security, finances, etc. It concludes with recommendations on which method would be more suitable to use, depending on the data and the problems in hand, in order to create efficient information technologies across application domains.