The reproduction of color across different media and viewing condition requires color-appearance modeling. In order to obtain an effective and efficient mapping of color appearance, the authors define here a method to approximate the combination of the forward and reverse Hunt94 color-appearance models. The method consists in learning by examples the Hunt94 color mapping for each desired experimental setup. Learning is done by feed-forward neural networks trained with the back-propagation algorithm on training sets derived from the ANSI IT8 7.2 color target. Experimental results confirm the feasibility of the method.