We live in an age of networks. From the social graph of Facebook to the interactions of proteins in the body, more and more of the world is being conceived of and represented as the connections in a network. And understanding of those connections can sometimes have stunning business implications, as when Larry Page and Sergey Brin of Stanford University first proposed modeled networks of Web pages, called "PageRank," the foundation of Google. Some heavy hitters in artificial intelligence have been working on ways to make machine learning techniques smarter about understanding networks. Late last week, a group of those researchers reported progress in having a neural network figure out the structure of a various networks without having full knowledge of all of a network.