A2BCD: Asynchronous Acceleration with Optimal Complexity

Authors: Robert Hannah, Fei Feng, Wotao Yin

ICLR 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental To investigate the performance of A2BCD, we solve the ridge regression problem. ... We use the datasets w1a (47272 samples, 300 features), wxa ... and aloi ... We use 40 threads on two 2.5GHz 10-core Intel Xeon E5-2670v2 processors. ... In Table 5, we plot the sub-optimality vs. time for decreasing values of λ, which corresponds to increasingly large condition numbers κ.
Researcher Affiliation Academia Robert Hannah , Fei Feng , Wotao Yin Department of Mathematics University of California, Los Angeles 520 Portola Plaza, Los Angeles, CA 90095, USA
Pseudocode Yes Algorithm 1 Shared-memory implementation of A2BCD
Open Source Code No The paper mentions implementation details ('implemented in a multi-threaded fashion using C++11 and GNU Scientific Library') but does not provide an explicit statement or link indicating that the source code for their method is publicly available.
Open Datasets Yes We use the datasets w1a (47272 samples, 300 features), wxa which combines the data from from w1a to w8a (293201 samples, 300 features), and aloi (108000 samples, 128 features) from LIBSVM Chang & Lin (2011).
Dataset Splits No The paper mentions the datasets and their sizes but does not specify the training, validation, or test split percentages or sample counts for reproduction.
Hardware Specification Yes We use 40 threads on two 2.5GHz 10-core Intel Xeon E5-2670v2 processors.
Software Dependencies Yes The algorithm is implemented in a multi-threaded fashion using C++11 and GNU Scientific Library.
Experiment Setup Yes The parameters for each algorithm are tuned to give the fastest performance, so that a fair comparison is possible. ... Through simple tuning though, we found that ψ = 0.25 resulted in good performance.