Measure Estimation in the Barycentric Coding Model

Authors: Matthew Werenski, Ruijie Jiang, Abiy Tasissa, Shuchin Aeron, James M Murphy

ICML 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, we demonstrate the utility of the BCM and associated estimation procedures in three application areas: (i) covariance estimation for Gaussian measures; (ii) image processing; and (iii) natural language processing.
Researcher Affiliation Academia 1Department of Computer Science, Tufts University 2Department of Electrical and Computer Engineering, Tufts University 3Department of Mathematics, Tufts University
Pseudocode Yes Algorithm 1 Estimate λ; Algorithm 2 Estimate λ on Point Clouds
Open Source Code Yes Code to reproduce results is available at https://github.com/MattWerenski/BCM
Open Datasets Yes taking as experimental data the MNIST dataset of 28 × 28 pixel images of hand-written digits (Le Cun, 1998). ... We use the publicly available dataset provided by (Huang et al., 2016).
Dataset Splits No The paper discusses training data and test data, but no explicit mention of a separate validation dataset or split is made.
Hardware Specification Yes All reported times are obtained using 20 Intel Xeon CPU E5-2660 V4 cores at 2.00 GHz.
Software Dependencies No The paper mentions 'PyTorch' but does not provide its version number or versions for other key software dependencies.
Experiment Setup Yes We set p = 6, d = 10, λ ∼ Unif( p), and µi ∼ Wish(Id) + 0.5Id with results averaged over 250 trials... To generate these figures we randomly sample k reference documents per topic and a test set of 100 random documents. We apply the four methods listed above and compute the accuracy on the test set. This procedure is repeated 50 times for each choice of k and the average accuracy is plotted.