Abstract

In this workshop, we want to advance theory, methods and tools for allowing experts to express prior coded knowledge for automatic data annotations that can be used to train arbitrary deep neural networks for prediction. The ICLR 2021 Workshop on Weak Supervision aims at advancing methods that help modern machine-learning methods to generalize from knowledge provided by experts, in interaction with observable (unlabeled) data. We called for both long and short papers and received 26 submissions, all of which were double-blindly reviewed by a pool of 29 reviewers. In total, 15 papers were accepted. All the accepted contributions are listed in these Proceedings and those submitted as archival are included in full text. Learning with weak supervision is both studied from a theoretical perspective as well as applied to a variety of tasks from areas like natural language processing and computer vision. Therefore, the workshop brought together researchers from a wide range of fields, also bridging innovations from academia and the requirements of industry settings. The program of the workshop, besides 3 oral paper presentations and 12 posters in 2 poster sessions, included invited talks by Marine Carpuat, Heng Ji, Lu Jiang, Dan Roth and Paroma Varma. It closed with a panel discussion with the invited speakers. Snorkel AI provided funding to sponsor ICLR registrations to increase diversity.