Distributed Box-Constrained Quadratic Optimization for Dual Linear SVM

Authors: Ching-Pei Lee, Dan Roth

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

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments also show that our method is significantly faster than state-of-the-art distributed linear SVM algorithms including DSVM-AVE, Dis DCA and TRON. [...] 4. Experiments The following algorithms are compared in our experiments.
Researcher Affiliation Academia Ching-pei Lee LEECHINGPEI@GMAIL.COM Dan Roth DANR@ILLINOIS.EDU University of Illinois at Urbana-Champaign, 201 N. Goodwin Avenue, Urbana, IL 61801 USA
Pseudocode Yes Algorithm 1 A box-constrained quadratic optimization algorithm for distributedly solving (2)
Open Source Code Yes The code used in the experiments is available at http://github.com/leepei/distcd_exp/.
Open Datasets Yes The statistics of the data sets in our experiments are shown in Table 1. All of them are publicly available.3 [Footnote 3: http://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets.]
Dataset Splits Yes For webspam and url, test sets are not available so we randomly split the original data into 80%/20% as training set and test set, respectively.
Hardware Specification Yes We use 16 nodes in a cluster. Each node has two Intel HP X5650 2.66GHZ 6C Processors, and one core per node is used.
Software Dependencies Yes We use the package MPI-LIBLINEAR 1.96.2
Experiment Setup Yes We fix C = 1 in all experiments for a fair comparison in optimization. [...] In BQO-E and BQO-A, τ = 0.001 is used. [...] We follow Tseng & Yun (2009) to use β = 0.5, σ = 0.1, γ = 0.