Tehnici de inteligență artificială pentru afaceri / Artificial intelligence techniques for business

The thesis is dedicated to the application of artificial intelligence techniques for business, particularly for financial prediction. This work is focused on the following two research directions: credit scoring and financial time series prediction. Credit scoring is concerned with the prediction of financial risk and informing managerial decision making in the money business. The quality of risk analysis may affect the financial performance of the bank. For credit scoring, one has experimented a set of classifiers (Multilayer Perceptron (MLP), decision trees, Support Vector Machine (SVM), C4.5 algorithm cascaded with AdaBoost) as well as feature selection techniques (Principal Component Analysis (PCA) and ReliefF). On the other side, the financial time series prediction tools are useful to business leaders and organisations to improve decisions regarding the effects of future predicted changes. There have been considered the exchange rate forecasting and the stock market prediction. There have been chosen two neural network techniques for financial time series prediction: Nonlinear Autoregressive Exogenous (NARX) model and Deep Learning Long Short-Term Memory (LSTM) model. The research results have been published; five papers have been included in International Conference Proceedings (WOS indexed) and a sixth paper has been published in an International Journal (Scopus indexed).