Deep Learning

We propose a deep learning approach which incorporates prior domain knowledge into a principled probabilistic matrix factorization framework to enhance complain behavior predictions as well as their interpretability.