Asymptotic Guarantees for Learning Generative Models with the Sliced-Wasserstein Distance

Authors: Kimia Nadjahi, Alain Durmus, Umut Simsekli, Roland Badeau

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We illustrate the validity of our theory on both synthetic data and neural networks. ... We support our theory with experiments that are conducted on both synthetic and real data.
Researcher Affiliation Academia 1: LTCI, Télécom Paris, Institut Polytechnique de Paris, France 2: CMLA, ENS Cachan, CNRS, Université Paris-Saclay, France 3: Department of Statistics, University of Oxford, UK
Pseudocode No The paper does not include any pseudocode or algorithm blocks.
Open Source Code Yes We provide the code to reproduce the experiments.2 ... 2See https://github.com/kimiandj/min_swe.
Open Datasets Yes We use the MNIST dataset, made of 60 000 training images and 10 000 test images of size 28 28.
Dataset Splits No The paper mentions 60,000 training images and 10,000 test images for the MNIST dataset but does not specify a validation split or percentages for data partitioning.
Hardware Specification No The paper does not provide specific details about the hardware used for experiments, such as GPU/CPU models or memory specifications.
Software Dependencies No The paper mentions using the ADAM optimizer [32] but does not provide specific version numbers for any software components or libraries, which is required for reproducibility.
Experiment Setup Yes We trained for 20 000 iterations with the ADAM optimizer [32]. Our training objective is MESWE of order 2 approximated with 20 random projections and 20 different generated datasets. We design a neural network with the fully-connected configuration given in [16, Appendix D].