Prediction of the shear strength of reinforced masonry walls using a large experimental database and artificial neural networks