Co-Clustering for Multitask Learning
This work addresses the problem of improving multitask learning efficiency and performance for machine learning practitioners, though it appears incremental as it builds upon existing multitask methods.
The paper tackles multitask learning by introducing a framework that learns shared representations through task and feature co-clustering, resulting in a scalable algorithm that systematically outperforms existing state-of-the-art methods on synthetic and benchmark datasets.
This paper presents a new multitask learning framework that learns a shared representation among the tasks, incorporating both task and feature clusters. The jointly-induced clusters yield a shared latent subspace where task relationships are learned more effectively and more generally than in state-of-the-art multitask learning methods. The proposed general framework enables the derivation of more specific or restricted state-of-the-art multitask methods. The paper also proposes a highly-scalable multitask learning algorithm, based on the new framework, using conjugate gradient descent and generalized \textit{Sylvester equations}. Experimental results on synthetic and benchmark datasets show that the proposed method systematically outperforms several state-of-the-art multitask learning methods.