Deeply Learned Robust Matrix Completion for Large-scale Low-rank Data Recovery
This addresses robust data recovery for applications such as imaging and modeling, but appears incremental as it builds on existing matrix completion techniques with learnable parameters.
The paper tackles robust matrix completion for large-scale low-rank data recovery with missing entries and outliers, proposing a novel scalable non-convex method called LRMC that achieves linear convergence and superior performance in experiments on synthetic and real datasets like video background subtraction and satellite imagery.
Robust matrix completion (RMC) is a widely used machine learning tool that simultaneously tackles two critical issues in low-rank data analysis: missing data entries and extreme outliers. This paper proposes a novel scalable and learnable non-convex approach, coined Learned Robust Matrix Completion (LRMC), for large-scale RMC problems. LRMC enjoys low computational complexity with linear convergence. Motivated by the proposed theorem, the free parameters of LRMC can be effectively learned via deep unfolding to achieve optimum performance. Furthermore, this paper proposes a flexible feedforward-recurrent-mixed neural network framework that extends deep unfolding from fix-number iterations to infinite iterations. The superior empirical performance of LRMC is verified with extensive experiments against state-of-the-art on synthetic datasets and real applications, including video background subtraction, ultrasound imaging, face modeling, and cloud removal from satellite imagery.