Spectral k-Support Norm Regularization

Authors: Andrew M McDonald, Massimiliano Pontil, Dimitris Stamos

NeurIPS 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We apply the norms to real and synthetic matrix completion datasets. Our findings indicate that spectral k-support norm regularization gives state of the art performance, consistently improving over trace norm regularization and the matrix elastic net.
Researcher Affiliation Academia Andrew M. Mc Donald, Massimiliano Pontil, Dimitris Stamos Department of Computer Science University College London {a.mcdonald,m.pontil,d.stamos}@cs.ucl.ac.uk
Pseudocode Yes Algorithm 1 Computation of x = prox λ
Open Source Code No The paper does not provide an explicit statement or link for open-source code availability for the methodology described.
Open Datasets Yes The datasets we considered were Movie Lens 100k and Movie Lens 1M (http://grouplens.org/datasets/movielens/), which consist of user ratings of movies, and Jester 1 and Jester 3 (http://goldberg.berkeley.edu/jesterdata/), which consist of users and ratings of jokes (Jester 2 showed essentially identical performance to Jester 1).
Dataset Splits Yes we sampled uniformly a percentage ∈ {10%, 20%, 30%} of the entries for training, and used a fixed 10% for validation.
Hardware Specification No The paper does not provide any specific hardware details for the experimental setup.
Software Dependencies No The paper does not provide specific software dependencies with version numbers.
Experiment Setup Yes For the optimization we used an accelerated proximal gradient method (FISTA), see e.g. [12, 21, 22], with the percentage change in objective as convergence criterion, with a tolerance of 10^-5 for the simulated datasets and 10^-3 for the real datasets.