Bert Huang

Adversarial Labeling for Learning without Labels

2018 · arXiv (Cornell University)

We consider the task of training classifiers without labels. We propose a weakly supervised method---adversarial label learning---that trains classifiers to perform well against an adversary that chooses labels for training data. The weak supervision constrains …

Block Belief Propagation for Parameter Learning in Markov Random Fields

2018 · arXiv (Cornell University)

Traditional learning methods for training Markov random fields require doing inference over all variables to compute the likelihood gradient. The iteration complexity for those methods therefore scales with the size of the graphical models. In …