Sparse and Low-Rank Tensor Decomposition

Authors: Parikshit Shah, Nikhil Rao, Gongguo Tang

NeurIPS 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We validate our algorithm with numerical experiments.
Researcher Affiliation Collaboration Parikshit Shah parikshit@yahoo-inc.com Nikhil Rao nikhilr@cs.utexas.edu Gongguo Tang gtang@mines.edu
Pseudocode Yes Algorithm 1 Algorithm for sparse and low rank tensor decomposition
Open Source Code No The paper does not provide any statement or link indicating the release of source code for the described methodology.
Open Datasets No A tensor Z is generated as the sum of a low rank tensor X and a sparse tensor Y . The low-rank component is generated as follows: Three sets of r unit vecots ui, vi, wi R50 are generated randomly, independently and uniformly distributed on the unit sphere. Also a random positive scale factor (uniformly distributed on [0, 1] is chosen and the tensor X = Pr i=1 λi ui vi wi. The tensor Y is generated by (Bernoulli) randomly sampling its entries with probability p.
Dataset Splits No The paper describes generating synthetic data for experiments but does not specify explicit train/validation/test splits with percentages or sample counts.
Hardware Specification No The paper does not provide specific details about the hardware (e.g., CPU, GPU models, memory) used to run the experiments.
Software Dependencies No The optimization problem (6) is solved using CVX in MATLAB. No version numbers are provided for CVX or MATLAB.
Experiment Setup Yes In all our experiments, the regularization parameter was picked to be ν = 1 n. ... For each such p, we perform 10 trials... We run 5 independent trials