Large-Scale Matrix Factorization with Missing Data under Additional Constraints

Abstract

missing-data-matrix-factorization-computational-imaging-lab-iit-madras

Description

Results

large-scale-matrix-factorization-missing-data-main-result-computational-imaging-lab-iit-madras

Figure 1: Reconstruction rate vs. fraction of revealed entries, and timing results, for 500×500 matrices ➯ MF-LRSDP matches or beats OptSpace, alternation, and damped Newton while taking the least time.

matrix-factorization-real-dataset-computational-imaging-lab-iit-madras

Figure 3. Cumulative histogram of RMS pixel error over 25 trials on the Dinosaur, Giraffe, and Face sequences.

matrix-factorization-constraint-effect-computational-imaging-lab-iit-madras

Figure 4. Input point tracks vs. reconstructed tracks with and without additional constraints on the Dinosaur turntable sequence.

matrix-factorization-runtime-plot-computational-imaging-lab-iit-madras

Figure 2. Reconstruction rate vs. fraction of revealed entries for rank-5 square matrices of varying size and rank.

Visualizations

mf-lrsdp-sfm-reconstruction-animation-computational-imaging-lab-iit-madras

Animation: Structure from Motion reconstruction progressing across optimization stages ➯ front, side, and top views of the recovered 3D points along with reprojection error, bundle error, and confidence metrics as they converge.

mf-lrsdp-sparsity-impact-animation-computational-imaging-lab-iit-madras

Animation: Effect of data sparsity on reconstruction ➯ the observation matrix at a given missing-data percentage alongside the resulting reconstruction error as sparsity increases.

Publications