BanditMTL: Bandit-based Multi-task Learning for Text Classification

Task variance regularization, which can be used to improve the generalization of Multitask Learning (MTL) models, remains unexplored in multi-task text classification. Accordingly, to fill this gap, this paper investigates how the task might be effectively regularized, and consequently proposes a multi-task learning method based on adversarial multiarmed bandit. The proposed method, named BanditMTL, regularizes the task variance by means of a mirror gradient ascent-descent algorithm. Adopting BanditMTL in the multitask text classification context is found to achieve state-of-the-art performance. The results of extensive experiments back up our theoretical analysis and validate the superiority of our proposals.