Abstract: Robust principal component analysis (RPCA) is a critical tool in modern machine learning, which detects outliers in the task of low-rank matrix reconstruction. In this talk, we explore the non-convex approaches for large-scale RPCA problems. In particular, we study two high-efficiency methods: (1) Riemannian accelerated alternating projections; and (2) Robust CUR decompositions. The theoretical guarantees have been established for both methods. The numerical advantages of the proposed algorithms are verified on synthetic and real-world datasets.
https://univienna.zoom.us/j/61759769704?pwd=aGlMQlhKSFZacHh0NU5NazlSRUx0UT09