Asynchronous Parallel Greedy Coordinate Descent
Authors: Yang You, Xiangru Lian, Ji Liu, Hsiang-Fu Yu, Inderjit S. Dhillon, James Demmel, Cho-Jui Hsieh
NeurIPS 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In the first part of the paper, we analyze the theoretical behavior of Asy-GCD and prove a linear convergence rate. In the second part, we develop an efficient kernel SVM solver based on Asy-GCD in the shared memory multi-core setting. Since our algorithm is fully asynchronous each core does not need to idle and wait for the other cores the resulting algorithm enjoys good speedup and outperforms existing multi-core kernel SVM solvers including asynchronous stochastic coordinate descent and multi-core LIBSVM. |
| Researcher Affiliation | Academia | Yang You , + Xiang Ru Lian , + Ji Liu Hsiang-Fu Yu Inderjit S. Dhillon James Demmel Cho-Jui Hsieh + equally contributed University of California, Davis University of Rochester University of Texas, Austin University of California, Berkeley youyang@cs.berkeley.edu, xiangru@yandex.com, jliu@cs.rochester.edu {rofuyu,inderjit}@cs.utexas.edu, demmel@eecs.berkeley.edu chohsieh@cs.ucdavis.edu |
| Pseudocode | Yes | Algorithm 1 Asynchronous Parallel Greedy Coordinate Descent (Asy-GCD) ... Algorithm 2 Greedy Coordinate Descent (GCD) for Dual Kernel SVM ... Algorithm 3 Asy-GCD for Dual Kernel SVM |
| Open Source Code | No | The paper does not provide concrete access to source code (specific repository link, explicit code release statement, or code in supplementary materials) for the methodology described. |
| Open Datasets | Yes | Table 1: Data statistics. is number of training samples, d is dimensionality, t is number of testing samples. ijcnn1 49,990 91,701 22 32 2 covtype 464,810 116,202 54 32 32 webspam 280,000 70,000 254 8 32 |
| Dataset Splits | Yes | We follow the parameter settings in [12], where C and γ are selected by cross validation. |
| Hardware Specification | Yes | All the experiments are run on the same system with 20 CPUs and 256GB memory, where the CPU has two sockets, each with 10 cores. |
| Software Dependencies | No | The paper mentions implementation in C++ with Open MP, but does not provide specific version numbers for these software components to ensure reproducibility. |
| Experiment Setup | Yes | We follow the parameter settings in [12], where C and γ are selected by cross validation. All the experiments are run on the same system with 20 CPUs and 256GB memory... We locate 64GB for kernel caching for all the algorithms. In our algorithm, the 64GB will distribute to each core; for example, for Asy-GCD with 20 cores, each core will have 3.2GB kernel cache. ... we partition the data by running kmeans algorithm on a subset of 20000 training samples. |