One of the crucial tasks in many inference problems is the extraction of an underlying sparse graphical model from a given number of high-dimensional measurements. In machine learning, this is frequently achieved using, as a penalty term, the L p norm of the model parameters, with p≤ 1 for efficient dilution. Here we propose a statistical mechanics analysis of the problem in the setting of perceptron memorization and generalization. Using a replica approach, we are able to evaluate the relative performance of naive dilution (obtained by learning without dilution, following by applying a threshold to the model parameters), L 1 dilution (which is frequently used in convex optimization) and L 0 dilution (which is optimal but computationally hard to implement). Whereas both L p diluted approaches clearly outperform the naive approach, we find a small region where L 0 works almost perfectly and strongly outperforms the simpler to implement L 1 dilution.