Optimal Iterative Sketching Methods with the Subsampled Randomized Hadamard Transform

Authors: Jonathan Lacotte, Sifan Liu, Edgar Dobriban, Mert Pilanci

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate the performance of the IHS with refreshed Haar/SRHT sketches against refreshed Gaussian sketches.
Researcher Affiliation Academia Jonathan Lacotte Department of Electrical Engineering Stanford University, Sifan Liu Department of Statistics Stanford University, Edgar Dobriban Department of Statistics University of Pennsylvania, Mert Pilanci Department of Electrical Engineering Stanford University
Pseudocode No The paper does not contain any pseudocode or clearly labeled algorithm blocks.
Open Source Code No The paper does not provide any explicit statements about open-source code availability or links to code repositories for the described methodology.
Open Datasets Yes Second, we carry out a similar experiment with the CIFAR10 dataset, for which we consider one-vs-all classification.
Dataset Splits No The paper mentions using a 'synthetic data matrix' and 'CIFAR10 dataset', but does not explicitly state specific training, validation, or test splits by percentages, counts, or by referencing predefined splits with citations for reproducibility.
Hardware Specification No The paper does not provide specific details about the hardware (e.g., CPU/GPU models, memory) used for running the experiments.
Software Dependencies No The paper does not list specific software dependencies with version numbers used for the experiments.
Experiment Setup Yes For the SRHT, we use the step size µt = θ1,h/θ2,h prescribed in Theorem 4.1, where we replace ξ and γ by their finite sample approximations ξ m n.