A comparative survey of three AI techniques (NN, PSO, and GA) in financial domain

In the real world, the behaviors of financial applications are unstable and they change from time to time. Accordingly, dealing with such issues as nonlinear and time variant problems has been a serious problem in recent years. These types of problems along with inefficiency of the traditional models led to an increasing interest in artificial intelligence approaches. In this study, we briefly review three popular artificial intelligence methods, i.e., Artificial Neural Networks, Genetic Algorithms, and Particle Swarm Optimization, and compare their applications in financial domain. By considering the broad domain of financial applications, we classify financial market into three domains, including financial forecasting, credit evaluation, and portfolio management. For each technique, we have attempted to take the most recent and popular studies into account. The results are promising and represent that in handling financial problems, the performance and accuracy of the above mentioned artificial intelligence techniques are considerably higher, compared to the traditional statistical techniques, particularly in nonlinear models. Nevertheless, this superiority is not true in all cases.