Pseudo-task Augmentation: From Deep Multitask Learning to Intratask Sharing---and Back

摘要: Deep multitask learning boosts performance by sharing learned structure across related tasks. This paper adapts ideas from deep multitask learning to the setting where only a …