Given some partial information about a system we like to reconstruct the interaction pattern of its elements. These elements can be neurons in a neural network, genes in a cell, pixels in an image or computers in the Internet. In the simplest scenario we assume that elements take only two states and interact pairwise. In this talk I will introduce some methods that people use to deal with this problem.
 William Bialek et. al, Faster solutions of the inverse pairwise Ising problem, 2007.
 M. Mezard and T. Mora, Constraint satisfaction problems and neural networks: a statistical physics perspective, 2008.