Learning Diffusion Priors from Observations by Expectation Maximization

Authors: François Rozet, Gerome Andry, Francois Lanusse, Gilles Louppe

NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We conduct three experiments to demonstrate the effectiveness of our method. We design the first experiment around a low-dimensional latent variable x whose ground-truth distribution p(x) is known. ... The remaining experiments target two benchmarks from previous studies: corrupted CIFAR-10 and accelerated MRI.
Researcher Affiliation Academia François Rozet University of Liège francois.rozet@uliege.be Gérôme Andry University of Liège gandry@uliege.be François Lanusse Université Paris-Saclay, Université Paris Cité, CEA, CNRS, AIM francois.lanusse@cnrs.fr Gilles Louppe University of Liège g.louppe@uliege.be
Pseudocode Yes We summarize the pipeline in Algorithms 1, 2 and 3, provided in Appendix A due to space constraints.
Open Source Code Yes The code for all experiments is made available at https://github.com/francois-rozet/ diffusion-priors.
Open Datasets Yes Following Daras et al. [80], we take the 50 000 training images of the CIFAR-10 [81] dataset as latent variables x.
Dataset Splits No The paper refers to using the training images from CIFAR-10 and fastMRI datasets but does not explicitly describe how these datasets were split into training, validation, and test sets for the experiments themselves, beyond using the 'training set' of existing benchmarks.
Hardware Specification Yes Each EM iteration (including sampling and training) takes around 4 h on 4 A100 (40GB) GPUs.
Software Dependencies No All experiments are implemented within the JAX [73] automatic differentiation framework. The paper mentions JAX but does not provide specific version numbers for JAX or any other key software dependencies like PyTorch, CUDA, or specific library versions used for implementation.
Experiment Setup Yes Experiment details are provided in Appendix C.