In this study, we test the ability of neural networks to determine the composition of magmatic rocks from their laboratory spectra. We first describe the structure and behaviour of the multilayer perceptron that we implement and train for quantitative characterization. For that purpose, reference laboratory spectra of mafic minerals from both natural and synthetic samples are used. As their composition in terms of the three mafic minerals, olivine (OL), orthopyroxene (OPX) and clinopyroxene (CPX) are known, those spectra are given as inputs during the learning phase of the neural network. In the analysis phase, we use the neural network to process spectra acquired on SNCs (Shergottites, Nakhlites, Chassignites) meteorite samples that are considered to be representative of Mars surface. The network outputs mineralogical compositions very quickly, performing only explicit operations. Our preliminary results show that neural networks are able to quantify mafic minerals, especially in the case of complex mixtures, with much improved computer efficiency and comparable accuracy compared to usual methods. This is very promising regarding future analysis of huge datasets.