Efficient Convex Completion of Coupled Tensors using Coupled Nuclear Norms
Authors: Kishan Wimalawarne, Hiroshi Mamitsuka
NeurIPS 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We carried out several simulations and real world data experiments to evaluate empirical performances of our proposed methods. Through simulation and real-data experiments, we demonstrate that proposed norms achieve better performance for coupled completion compared to existing coupled norms. |
| Researcher Affiliation | Academia | Kishan Wimalawarne1 and Hiroshi Mamitsuka1,2 1Bioinformatics Center, Kyoto University, Kyoto, Japan 2Department of Computer Science, Aalto University, Espoo, Finland |
| Pseudocode | Yes | We provide a Frank-Wolfe based optimization method to solve above completion models in the Section B of the Appendix. |
| Open Source Code | Yes | Code and data are available at http://kishan-wimalawarne.com/onewebmedia/Neur IPS_2018_code.rar |
| Open Datasets | Yes | We used the UCLAF dataset as our real data experiment. The UCLAF dataset (Zheng et al., 2010) is a commonly used benchmark dataset for coupled tensor completion (Ermis et al., 2015; Wimalawarne et al., 2018). |
| Dataset Splits | Yes | In order to generate datasets for simulations, we selected training sets of 30, 50, and 70 percentages from total number of elements of the tensor and the matrix, 10 percent as validation sets and the rest as test sets. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used for running experiments (e.g., GPU models, CPU specifications, memory). |
| Software Dependencies | No | The paper does not specify version numbers for any software dependencies, libraries, or frameworks used in the experiments. |
| Experiment Setup | Yes | For all these norms, we used the regularization parameters λ1, . . . , λ5 in the range from 0.01 to 50 with intervals of 1. As baseline methods, we performed completion of each individual tensor using the overlapped trace norm (OTN) and the scaled latent trace norm (SLTN) and individual matrix completion using the matrix trace norm (MTN). We also used the tensor nuclear norm as a baseline method to evaluate individual tensor completion. For all the baseline methods, we selected the optimal regularization parameters from the range of 0.01 to 5 in divisions of 0.025. |