Sliced Wasserstein Kernel for Persistence Diagrams

Authors: Mathieu Carrière, Marco Cuturi, Steve Oudot

ICML 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we compare k SW to k PSS and k PWG on several benchmark applications for which PDs have been proven useful. We compare these kernels in terms of classification accuracies and computational cost.
Researcher Affiliation Academia 1INRIA Saclay 2CREST, ENSAE, Universit e Paris Saclay.
Pseudocode Yes Algorithm 1 Computation of SWM
Open Source Code No The paper references LIBSVM: 'Software available at http://www.csie.ntu.edu. tw/ cjlin/libsvm.' but does not provide a direct link to the source code for the methodology described in this paper.
Open Datasets Yes We use some categories of the mesh segmentation benchmark of Chen et al. (Chen et al., 2009), which contains 3D shapes classified in several categories ( airplane , human , ant ...).
Dataset Splits Yes The cost factor C is cross-validated in the following grid: {0.001, 0.01, 0.1, 1, 10, 100, 1000}.
Hardware Specification Yes results are averaged over 10 runs on a 2.4GHz Intel Xeon E5530 Quad Core.
Software Dependencies No The paper mentions 'LIBSVM (Chang & Lin, 2011)' and 'The GUDHI Project (The GUDHI Project, 2015)' but does not specify version numbers for these software components.
Experiment Setup Yes The cost factor C is cross-validated in the following grid: {0.001, 0.01, 0.1, 1, 10, 100, 1000}.