Fast DPP Sampling for Nystrom with Application to Kernel Methods

Authors: Chengtao Li, Stefanie Jegelka, Suvrit Sra

ICML 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We present several empirical results that support our theoretical analysis, and demonstrate the superior performance of DPP-based landmark selection compared with existing approaches.
Researcher Affiliation Academia Chengtao Li CTLI@MIT.EDU Stefanie Jegelka STEFJE@MIT.EDU Suvrit Sra SUVRIT@MIT.EDU Massachusetts Institute of Technology
Pseudocode Yes Algorithm 1 Gibbs sampler for c-DPP
Open Source Code No No explicit statement providing concrete access to source code for the methodology described in this paper was found.
Open Datasets Yes We use 8 datasets: Abalone, Ailerons, Elevators, Comp Act, Comp Act(s), Bank32NH, Bank8FM and California Housing4. We subsample 4,000 points from each dataset (3,000 training and 1,000 test). 4http://www.dcc.fc.up.pt/ ltorgo/ Regression/Data Sets.html
Dataset Splits Yes We subsample 4,000 points from each dataset (3,000 training and 1,000 test)... Throughout our experiments, we use an RBF kernel and choose the bandwidth σ and regularization parameter λ for each dataset by 10fold cross-validation.
Hardware Specification No No specific hardware details (e.g., GPU/CPU models, processor types, memory amounts, or detailed computer specifications) used for running experiments were found.
Software Dependencies No No specific ancillary software details (e.g., library or solver names with version numbers) were found.
Experiment Setup Yes Throughout our experiments, we use an RBF kernel and choose the bandwidth σ and regularization parameter λ for each dataset by 10fold cross-validation. We initialize the Gibbs sampler via Kmeans++ and run for 3,000 iterations. Results are averaged over 3 random subsets of data.