Polynomial Preconditioning for Gradient Methods

Authors: Nikita Doikov, Anton Rodomanov

ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Numerical experiments confirm the efficiency of our preconditioning strategies for solving various machine learning problems.
Researcher Affiliation Academia Nikita Doikov 1 Anton Rodomanov 2 1EPFL, Switzerland 2UCLouvain, Belgium. Correspondence to: Nikita Doikov <nikita.doikov@epfl.ch>, Anton Rodomanov <anton.rodomanov@uclouvain.be>.
Pseudocode Yes Algorithm 1 Preconditioned Basic Gradient Method
Open Source Code No No explicit statement or link providing access to the source code for the methodology described in the paper was found.
Open Datasets Yes Figure 2: Leading eigenvalues (in the logarithmic scale) of the curvature matrix B, for several typical datasets2. 2https://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/
Dataset Splits No The paper does not provide specific details on training, validation, or test dataset splits.
Hardware Specification Yes Clock time was evaluated using the machine with Intel Core i5 CPU, 1.6GHz; 8 GB RAM. All methods were implemented in Python.
Software Dependencies No The paper states 'All methods were implemented in Python.' but does not provide specific version numbers for Python or any libraries used.
Experiment Setup No The paper discusses parameters and adaptive search procedures but does not provide specific numerical values for hyperparameters or other concrete training configurations for the experiments shown.