Multiplex networks arise in many real-world systems when different kinds of relationship connect the elements of the network. As a result, studying methods to detect communities in such network becomes increasingly important. Common approaches to this problem try to generalize methods from monoplex networks or apply tensor decomposition, but this field lacks of better procedures and novel methods. Chi and Kolda  presented a new tensor factorization method focused on minimizing the Kullback-Leibler divergence by using Poisson distribution, since the random variation found in sparse count data is better described by such distribution. Because networks have this characteristic of sparse data, this project aims to develop algorithms for treating community detection in multiplex networks applying tensor factorization, hoping the resulting factors will represent communities that emerge in multiplex networks.
 E. C. Chi and T. G. Kolda. On Tensors, Sparsity, and Nonnegative Factorizations, SIAM Journal on Matrix Analysis and Applications 33(4):1272-1299, December 2012. URL: http://arxiv.org/abs/1112.2414.