Nonlinear Acceleration of Stochastic Algorithms

Authors: Damien Scieur, Francis Bach, Alexandre d'Aspremont

NeurIPS 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, in Section 6 we describe numerical experiments which confirm the theoretical bounds and show the practical efficiency of this acceleration. 6 Numerical Experiments
Researcher Affiliation Academia Damien Scieur INRIA, ENS, PSL Research University, Paris France damien.scieur@inria.fr Francis Bach INRIA, ENS, PSL Research University, Paris France francis.bach@inria.fr Alexandre d Aspremont CNRS, ENS, PSL Research University, Paris France aspremon@ens.fr
Pseudocode Yes Algorithm 1 Regularized Nonlinear Acceleration (RNA)
Open Source Code No The paper does not explicitly state that the source code for the methodology is open-source, nor does it provide any links to a code repository.
Open Datasets Yes We compare these various methods for solving least-squares regression and logistic regression on several datasets (Table 1), with several condition numbers κ: well (κ = 100/N), moderately (κ = 1/N) and badly (κ = 1/100N) conditioned. In this section, we present the numerical results on Sid (Sido0 dataset, where N = 12678 and d = 4932) with bad conditioning, see Figure 2.
Dataset Splits No The paper describes the datasets used (e.g., Sido0) but does not provide explicit details on how these datasets were split into training, validation, or test sets, nor does it mention specific percentages, sample counts, or splitting methodologies.
Hardware Specification No The paper states 'The experiments are done in Matlab.' but does not provide any specific hardware details such as CPU models, GPU models, or memory specifications used for the experiments.
Software Dependencies No The paper states 'The experiments are done in Matlab.' but does not provide specific version numbers for Matlab or any other software dependencies, libraries, or solvers used.
Experiment Setup Yes RNA+SGD. The regularized nonlinear acceleration Algorithm 1 applied to a sequence of k iterates of SGD, with k = 10 and λ = RT R /10 6. The parameter λ is found by a grid search of size k, the size of the input sequence... We set k = 10 for all the experiments.