Coded Distributed Computing for Inverse Problems
Authors: Yaoqing Yang, Pulkit Grover, Soummya Kar
NeurIPS 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | For validation of our theory, we performed experiments to compare coded and replication-based computation for a graph mining problem, namely personalized Page Rank [18] using the classical power-iteration method [19]. We conduct experiments on the Twitter and Google Plus social networks under a deadline on computation time using a given number of workers on a real computation cluster (Section 6). |
| Researcher Affiliation | Academia | Yaoqing Yang, Pulkit Grover and Soummya Kar Carnegie Mellon University {yyaoqing, pgrover, soummyak}@andrew.cmu.edu |
| Pseudocode | Yes | Algorithm 1 Coded Distributed Linear Inverse |
| Open Source Code | No | The paper does not provide a link or explicit statement about the availability of open-source code for the methodology described. |
| Open Datasets | Yes | We test the performance of the coded linear inverse algorithm for the Page Rank problem on the Twitter graph and the Google Plus graph from the SNAP datasets [28]. |
| Dataset Splits | No | The paper describes experiments on full graphs (Twitter and Google Plus) but does not specify training, validation, or test data splits in terms of percentages or sample counts for model development and evaluation. |
| Hardware Specification | No | We use the HT-condor framework in a cluster to conduct the experiments. (No specific hardware details like GPU/CPU models are provided). |
| Software Dependencies | No | The paper does not specify any software dependencies with version numbers. |
| Experiment Setup | Yes | The task is to solve k = 100 personalized Page Rank problems in parallel using n = 120 workers. The coded Page Rank uses n workers to solve these k = 100 equations using Algorithm 1. We use a (120, 100) code where the generator matrix is the submatrix composed of the first 100 rows in a 120 × 120 DFT matrix. |