Faster Derivative-Free Stochastic Algorithm for Shared Memory Machines

Authors: Bin Gu, Zhouyuan Huo, Cheng Deng, Heng Huang

ICML 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The experimental results on the application of ensemble learning confirm that our Asy SZO+ has a faster convergence rate than the existing (asynchronous) stochastic zeroth-order algorithms.
Researcher Affiliation Academia 1Department of Electrical and Computer Engineering, University of Pittsburgh, PA, USA 2School of Electronic Engineering, Xidian University, Xi an, Shaanxi, China.
Pseudocode Yes Algorithm 1 Asynchronous Stochastic Zeroth-order Optimization (Asy SZO) and Algorithm 2 New Asynchronous Stochastic Zeroth-Order Algorithm with Variance Reduction and Mini-Batch (Asy SZO+)
Open Source Code No No explicit statement or link providing access to the source code for the described methodology was found. The paper only states 'We implement our Asy SZO+ in C++, where the shared memory parallel computation is handled via Open MP (Chandra, 2001). Similarly, we implement Asy SZO using C++ and Open MP.'
Open Datasets Yes As mentioned in (Dror et al., 2012), KDD-Cup 2011 challenged the community to identify user tastes in music by leveraging Yahoo! Music user ratings. ... We were able to obtain the predicted ratings of N = 237 individual models on the KDD-Cup Track 1 testing data set from the NTU KDD-Cup team, which is a matrix A with 6,005,940 rows (corresponding to the 6,005,940 samples in the testing data set) and 237 columns (corresponding to the outputs of 237 models on all samples in the testing set).
Dataset Splits No The paper mentions 'validation set' and its size ('l = 6, 005, 940 is the size of the validation set') and a 'testing data set', but does not provide specific details on how the dataset was split into training, validation, and testing portions (e.g., percentages or splitting methodology).
Hardware Specification Yes Our experiments are performed on a 32-core two-socket Intel Xeon E5-2699 machine where each socket has 16 cores.
Software Dependencies No The paper states 'We implement our Asy SZO+ in C++, where the shared memory parallel computation is handled via Open MP (Chandra, 2001),' but does not provide specific version numbers for these software components.
Experiment Setup No The paper mentions 'mini-batch sizes 1 and 100' and that 'γ is set as a fixed constant' but does not provide the specific value of γ or other detailed hyperparameters like learning rates, number of epochs, or optimizer settings used in the experiments.