Unsupervised Deep Haar Scattering on Graphs

Authors: Xu Chen, Xiuyuan Cheng, Stephane Mallat

NeurIPS 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental 4 Numerical ExperimentsThe performance of a Haar scattering classification is tested on scrambled images, whose graph geometry is unknown. Results are provided for MNIST and CIFAR-10 image data bases. Classification experiments are also performed on scrambled signals whose samples are on an irregular grid of a sphere.
Researcher Affiliation Academia 1Department of Electrical Engineering, Princeton University, NJ, USA 2D epartement d Informatique, Ecole Normale Sup erieure, Paris, France
Pseudocode No The paper describes the computational process through text and equations (e.g., equations 1-5 and Figure 1), but it does not include a formal pseudocode block or algorithm listing.
Open Source Code Yes All computations can be reproduced with a software available at www.di.ens.fr/data/scattering/haar.
Open Datasets Yes MNIST is a data basis with 6 × 10^4 hand-written digit images of size d = 210, with 5 × 10^4 images for training and 10^4 for testing. CIFAR-10 images are color images of 32 × 32 pixels... with a total of 5 × 10^4 training examples and 10^4 testing examples.
Dataset Splits No The paper states '5 × 10^4 images for training and 10^4 for testing' for MNIST and CIFAR-10 datasets, but does not explicitly describe a separate validation split or the methodology for such a split.
Hardware Specification No The paper does not specify any particular hardware components such as GPU models, CPU types, or memory specifications used for running the experiments.
Software Dependencies No The paper mentions that 'All computations can be reproduced with a software available at www.di.ens.fr/data/scattering/haar,' but it does not list specific software dependencies or their version numbers.
Experiment Setup Yes The scattering scale 2J d is the invariance scale. Scattering coefficients are computed up to the a maximum order m, which is set to 4 in all experiments. Indeed, higher order scattering coefficient have a negligible relative energy, which is below 1%. The unsupervised learning algorithm computes N multiresolution approximations, corresponding to N different scattering transforms... The supervised dimension reduction selects a final set of M orthogonalized scattering coefficients. We set M = 1000 in all numerical experiments.