Learning Equivariant Energy Based Models with Equivariant Stein Variational Gradient Descent

Authors: Priyank Jaini, Lars Holdijk, Max Welling

NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We empirically demonstrate the performance of equivariant EBMs using equivariant SVGD in 5. We consider real-world applications comprising of problems from many-body particle systems, molecular structure generation and, classification and generation for image datasets.
Researcher Affiliation Collaboration Priyank Jaini Bosch-Delta Lab University of Amsterdam Lars Holdijk Bosch-Delta Lab University of Amsterdam Max Welling Bosch-Delta Lab University of Amsterdam
Pseudocode Yes Algorithm 1: Equivariant EBM training
Open Source Code No The paper does not provide an explicit statement or link indicating the availability of open-source code for the described methodology.
Open Datasets Yes QM9: QM9 is a molecular dataset containing over 145,000 molecules used for moleccular property prediction. Fashion MNIST: In this experiment, we take the Fashion MNIST dataset with training set consisting regular images whereas the test set is processed to contain images that are randomly rotated using the C4-symmetry group.
Dataset Splits No The paper mentions a 'training set' and 'test set' for Fashion MNIST but does not provide specific details on validation splits or quantitative split percentages/counts for any dataset.
Hardware Specification No The paper does not provide any specific details about the hardware used for running the experiments.
Software Dependencies No The paper mentions software components like 'Adam optimizer' and specific neural network architectures (e.g., 'C4 steerable CNN', 'E-GNN') but does not provide specific version numbers for any software dependencies.
Experiment Setup Yes We keep the experimental setup i.e. number of particles and number iterations exactly the same for both vanilla SVGD and equivariant SVGD. ... We run vanilla SVGD with up to 32 times more particles than invariant SVGD. ... Update θ using θ and Adam optimizer