Randomized Dimensionality Reduction for Facility Location and Single-Linkage Clustering

Authors: Shyam Narayanan, Sandeep Silwal, Piotr Indyk, Or Zamir

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

Reproducibility Variable Result LLM Response
Research Type Experimental Lastly, we provide experimental results to validate the quality of solutions and the speedup due to the dimensionality reduction. ... Finally, we present an experimental evaluation of the algorithms suggested by our results.
Researcher Affiliation Academia 1Electrical Engineering and Computer Science Department, Massachusetts Institute of Technology, Cambridge, MA, USA 2Institute of Advanced Study, Princeton, NJ, USA.
Pseudocode No The paper refers to existing algorithms like the Mettu Plaxton (MP) algorithm and Boruvka algorithm but does not include any pseudocode or algorithm blocks for its own methods.
Open Source Code No The paper mentions the use of the 'mlpack machine learning library' (Curtin et al., 2018) but does not provide any statement or link indicating that its own methodology's source code is available.
Open Datasets Yes Faces Dataset: This dataset is used in the influential ISOMAP paper and consists of 698 images of faces in dimension 4096 (Tenenbaum et al., 2000). ... MNIST 2 Dataset: 1000 randomly chosen images from the MNIST dataset (dimension 784) restricted to the digit 2. We picked 2 since it is considered in the original ISOMAP paper (Tenenbaum et al., 2000).
Dataset Splits No The paper mentions using datasets for experiments but does not provide specific details on training, validation, or test splits (e.g., percentages, sample counts, or explicit standard split references).
Hardware Specification Yes All of our experiments were done on a CPU with i5 2.7 GHz dual core and 8 GB RAM.
Software Dependencies No The paper mentions the use of 'mlpack machine learning library' but does not provide a specific version number for it or other software dependencies.
Experiment Setup Yes We project our datasets and compute a facility location clustering with the opening costs scaled so that n/2, n/5, and n/10 facilities are opened respectively. ... All of our experimental results are averaged over 20 independent trials and the projection dimension d ranges from 5 to 20 inclusive.