estimating accuracies of classifiers using only unlabeled data
enhacing machine translation using contextual parameter generation and curriculum learning
enabling neural networks to better perform multi-task learning and more
type-safe linear algebra, tensors, and neural networks
editor and founding member for the CMU machine learning department blog
extracting a knowledge-base from the web, while constantly improving at it
a novel environment for experimenting with never-ending learning
active learning amidst logical knowledge
exploring how to more efficiently teach neural networks
