This talk focuses on the model based identification of network dynamics from steady state sample observations. We study two types of network dynamics that have been used as qualitative models for opinion dynamics in social networks and for gene regulatory networks. The dynamics models are described in
a generic form and their corresponding steady states are characterized. 
In contrast to the common assumption with full rank input necessary for methods such as Graph LASSO, we show that model based identification  
works when network data are low-rank, which is more typical and that the
proposed methodology offers superior performance compared to state-of-the-art methods. In addition, we are able to identify directed graph dependency, which correlation alone cannot capture.