A novel variational form of the Schatten-$p$ quasi-norm

Authors: Paris Giampouras, Rene Vidal, Athanasios Rontogiannis, Benjamin Haeffele

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, the efficiency of our approach is empirically shown on a matrix completion problem.In this section we provide simulated and real data experimental results that advocate the merits of the proposed variational form of the Sp quasi-norm in the case of the matrix completion problem.
Researcher Affiliation Academia Paris Giampouras Mathematical Institute for Data Science Johns Hopkins University parisg@jhu.edu René Vidal Mathematical Institute for Data Science Johns Hopkins University rvidal@jhu.edu Athanasios Rontogiannis IAASARS National Observatory of Athens tronto@noa.gr Benjamin D. Haeffele Mathematical Institute for Data Science Johns Hopkins University bhaeffele@jhu.edu
Pseudocode No The main paper text does not contain pseudocode or a clearly labeled algorithm block. It mentions 'Analytical details of the algorithm are provided in the supplement,' but the supplement is not included in the provided text.
Open Source Code No The paper does not provide concrete access to source code, such as a specific repository link or an explicit code release statement.
Open Datasets Yes Real data. We next test the algorithms on the Movie Lens-100K dataset, [24] which contains 100,000 ratings (integer values from 1 to 5) for 943 movies by 1682 users. [24] Movielens dataset. [Online]. Available: https://grouplens.org/datasets/movielens/
Dataset Splits No The paper mentions 'missing rates' for the observed entries but does not specify explicit training, validation, or test dataset splits, or cross-validation setup.
Hardware Specification Yes All experiments are conducted on a Mac Book Pro with 2.6 GHz 6-Core Intel Core i7 CPU and 16GB RAM using MATLAB R2019b.
Software Dependencies Yes All experiments are conducted on a Mac Book Pro with 2.6 GHz 6-Core Intel Core i7 CPU and 16GB RAM using MATLAB R2019b.
Experiment Setup Yes For each case we initialize all matrix factorization based algorithms with ranks ranging from 10 to 50 with step-size 5.