Dimensionality Reduction for the Sum-of-Distances Metric

Authors: Zhili Feng, Praneeth Kacham, David Woodruff

ICML 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We perform experiments to empirically verify that we can attain a non-trivial amount of data reduction while still being able to compute an approximate sum of distances to a kdimensional shape. In our experiments, we set n = 10000 and k = 5. We use various subspaces to compute an approximation to the sum of distances to a k center set.
Researcher Affiliation Academia 1 Carnegie Mellon University, Pittsburgh, USA.
Pseudocode Yes Algorithm 1 POLYAPPROX; Algorithm 2 EPSAPPROX; Algorithm 3 DIMENSIONREDUCTION; Algorithm 4 COMPLETEDIMREDUCE
Open Source Code Yes An implementation of our Algorithm 3 and code for our experiments is available at here2. https://gitlab.com/praneeth10/ dimensionality-reduction-for-sum-of-distances
Open Datasets Yes We run our dimensionality reduction algorithm on a randomly chosen subset A of size 10000 of the Cover Type dataset (Dua and Graff, 2017). URL http://archive.ics.uci.edu/ml.
Dataset Splits No The paper mentions using a subset of the Cover Type dataset but does not specify any training, validation, or test splits, or the use of cross-validation.
Hardware Specification No The paper does not provide any specific details about the hardware used to run the experiments (e.g., CPU, GPU models, memory).
Software Dependencies No The paper does not provide specific version numbers for any software dependencies used in the experiments (e.g., libraries, frameworks).
Experiment Setup No The paper mentions setting a "target dimension of 100" in the experiments, but it does not provide comprehensive details on other experimental setup parameters like learning rates, batch sizes, optimizers, or training schedules.