Unambiguous reconstruction of network structure using avalanche dynamics

A robust method for inferring the structure of networks is presented based on the one-to-one correspondence between the expected composition of cascades of bursts of activity, called crackling noise or avalanches, and the weight matrix. Using a model of neuronal avalanches as a paradigmatic example, we derive this correspondence exactly by calculating the closed-form expression of the joint probability distribution of avalanche sizes obtained by counting separately the number of elements active in each subnetwork during avalanches.