We present two adaptive predistortion techniques for non-linear satellite amplifiers. The amplifier gives two kind of distortions: amplitude distortion (AM/AM conversion) and phase distortion (AM/PM conversion). We compare the performance of these predistortion techniques applied to the same amplifier. Both predistortions are realized with multilayer perceptron (MLP) neural networks (NN) associated with the backpropagation (BP) algorithm. The first adaptive predistortion technique identifies the inverse AM/AM conversion with one NN and the direct AM/PM conversion with another one. The second one realizes directly the predistortion by training an input NN at the input of the amplifier. This NN contains one structure for the inversion of the AM/AM conversion and one for the AM/PM conversion.