PROJECT TITLE :

Multi-Target Regression via Robust Low-Rank Learning - 2017

ABSTRACT:

Multi-target regression has recently regained great popularity thanks to its capability of simultaneously learning multiple relevant regression tasks and its wide applications in information mining, computer vision and medical image analysis, whereas nice challenges arise from jointly handling inter-target correlations and input-output relationships. During this paper, we have a tendency to propose Multilayer Multi-target Regression (MMR) that enables simultaneously modeling intrinsic inter-target correlations and nonlinear input-output relationships during a general framework via robust low-rank learning. Specifically, the MMR will explicitly encode intertarget correlations in a very structure matrix by matrix elastic nets (MEN); the MMR can work along with the kernel trick to effectively disentangle highly complex nonlinear input-output relationships; the MMR will be efficiently solved by a replacement alternating optimization algorithm with guaranteed convergence. The MMR leverages the strength of kernel strategies for nonlinear feature learning and also the structural advantage of multi-layer learning architectures for inter-target correlation modeling. Additional importantly, it offers a replacement multi-layer learning paradigm for multi-target regression that is endowed with high generality, flexibility and expressive ability. Intensive experimental analysis on 18 diverse real-world datasets demonstrates that our MMR can achieve consistently high performance and outperforms representative state-of-the-art algorithms, which shows its nice effectiveness and generality for multivariate prediction.