Randomized Approach to Matrix Completion: Applications in Recommendation Systems and Image Inpainting

Abstract

We present a novel method for matrix completion, specifically designed for matrices where one dimension is significantly larger than the other. Our Columns Selected Matrix Completion (CSMC) method combines Column Subset Selection with Low-Rank Matrix Completion to efficiently reconstruct incomplete datasets. CSMC substantially reduces computational cost while preserving the solution quality of state-of-the-art convex matrix completion techniques. Each stage of CSMC involves solving a convex optimization problem. We introduce two algorithmic implementations of CSMC, each tailored to problems of different scales. A formal analysis is provided, outlining the necessary assumptions and the probability of exact recovery. To evaluate the impact of matrix size, rank, and the ratio of missing entries on solution quality and runtime, we conducted experiments on synthetic data. The method was also applied to two real-world tasks: recommendation systems and image inpainting. Our results show that CSMC achieves substantial runtime improvements while maintaining competitive accuracy compared to leading convex optimization-based methods.

References in corpus (7)

• Array Programming with NumPy
• Guaranteed Minimum-Rank Solutions of Linear Matrix Equations via Nuclear Norm Minimization
• Sketching as a Tool for Numerical Linear Algebra
• Robust CUR Decomposition: Theory and Imaging Applications
• Structured Gradient Descent for Fast Robust Low-Rank Hankel Matrix Completion
• Matrix Completion with Cross-Concentrated Sampling: Bridging Uniform Sampling and CUR Sampling
• Concentration properties of fractional posterior in 1-bit matrix completion