Subspace Robust Wasserstein Distances
Authors: François-Pierre Paty, Marco Cuturi
ICML 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We propose two algorithms to compute the latter formulation using entropic regularization, and illustrate the interest of this approach empirically. ... We conclude the paper with experiments in 6 to validate and illustrate our claims, on both simulated and real datasets. |
| Researcher Affiliation | Collaboration | 1CREST-ENSAE, Palaiseau, France 2Google Brain, Paris, France. |
| Pseudocode | Yes | Algorithm 1 Projected supergradient method for SRW ... Algorithm 2 Frank-Wolfe algorithm for regularized SRW |
| Open Source Code | No | The paper does not provide any explicit statements about releasing source code or links to a code repository for the described methodology. |
| Open Datasets | No | We consider the scripts of seven movies. Each script is transformed into a list of words, and using word2vec (Mikolov et al., 2018), into a measure over R300 where the weights correspond to the frequency of the words. (This does not provide concrete access to the movie script data itself or the specific word2vec model used for reproducibility.) |
| Dataset Splits | No | The paper describes experiments with simulated and real data (movie scripts), but it does not specify explicit train/validation/test splits (e.g., percentages, sample counts, or predefined splits) for these datasets. |
| Hardware Specification | No | The paper mentions 'on GPU' for computation time experiments but does not provide specific details such as GPU model, CPU, or memory specifications. |
| Software Dependencies | No | The paper mentions algorithms and tools like 'Sinkhorn algorithm' and 'word2vec (Mikolov et al., 2018)' but does not provide specific version numbers for any software dependencies. |
| Experiment Setup | Yes | Using k = 2 and Algorithm 2 with γ = 0.1 and stopping threshold ϵ = 0.05, we plot in Figure 8 the mean computation time of both SRW and Wasserstein distances on GPU, over 100 random samplings with n = 100. |