The project actually explored an important unsolved problem in deep learning called “unsupervised learning.” Almost every deep-learning product in commercial use today uses “supervised learning,” meaning that the neural net is trained with labeled data (like the images assembled by ImageNet). With “unsupervised learning,” by contrast, a neural net is shown unlabeled data and asked simply to look for recurring patterns. Researchers would love to master unsupervised learning one day because then machines could teach themselves about the world from vast stores of data that are unusable today—making sense of the world almost totally on their own, like infants.
When a group of district superintendents in West Michigan gathered for their regular quarterly meeting a few years ago, they bucked their regular meeting format and began discussing the needs of the nearly 5,000 third-graders in their districts who were demonstrating below-proficiency achievement in reading. What followed was the formation of a ground-breaking reading network among these districts, and formal commitments to support principals and teachers with literacy coaching, professional learning opportunities, and research field studies, among several other resolutions. The result: In just three years of focusing on the reading proficiency of at-risk students, member districts quadrupled the margin by which these students exceeded the state average in third-grade literacy.