Are exponential random graph models based on Maximum Entropy Principle?

Can anyone shed light on whether or not ERGMs are network models based on the maximum entropy principle? Especially if we say that the available information about the network is in the form of network measures, that is, we know that that graph in question G has $x_1(G), x_2(G),...,x_k(G)$ where $x_i$ is a measure, such as, the number of edges, number of wedges, number of triangles, etc.