A Randomized Algorithm to Reduce the Support of Discrete Measures

Authors: Francesco Cosentino, Harald Oberhauser, Alessandro Abate

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

Reproducibility Variable Result LLM Response
Research Type Experimental We give a simple geometric characterization of barycenters via negative cones and derive a randomized algorithm that computes this new measure by greedy geometric sampling . We then study its properties, and benchmark it on synthetic and real-world data to show that it can be very beneficial in the N n regime.
Researcher Affiliation Academia Francesco Cosentino Mathematical Institute University of Oxford The Alan Turing Institute francesco.cosentino@maths.ox.ac.uk Harald Oberhauser Mathematical Institute University of Oxford oberhauser@maths.ox.ac.uk Alessandro Abate Dept. of Computer Science University of Oxford The Alan Turing Institute alessandro.abate@cs.ox.ac.uk
Pseudocode Yes Algorithm 1 Basic measure reduction algorithm
Open Source Code Yes A Python implementation is available at https://github.com/Fra Cose/Recombination_Random_Algos.
Open Datasets Yes (i) 3D Road Network [21] that contains 434874 records and use the two attributes longitude and latitude to predict height, (ii) Household power consumption [22] that contains 2075259 records and use the two 2 attributes active and reactive power to predict voltage.
Dataset Splits No The paper discusses train, validation, and test in the schema context but does not provide specific splits (e.g., percentages or counts) for these datasets within the paper.
Hardware Specification Yes All experiments have been run on a Mac Book Pro, CPU: i7-7920HQ, RAM: 16 GB, 2133 MHz LPDDR3.
Software Dependencies No The paper mentions a "Python implementation" but does not specify version numbers for Python or any associated libraries/frameworks.
Experiment Setup No The paper describes the algorithms and their complexities but does not provide specific experimental setup details such as hyperparameters (e.g., learning rates, batch sizes, number of epochs) for training or optimization beyond the algorithmic steps themselves.