Scalable Probabilistic Model Selection For Network Representation Learning In Biological Network Inference