2014
NeurIPS
NeurIPS 2014
Unsupervised Deep Haar Scattering on Graphs
Abstract
The classification of high-dimensional data defined on graphs is particularly difficult when the graph geometry is unknown. We introduce a Haar scattering transform on graphs, which computes invariant signal descriptors. It is implemented with a deep cascade of additions, subtractions and absolute values, which iteratively compute orthogonal Haar wavelet transforms. Multiscale neighborhoods of unknown graphs are estimated by minimizing an average total variation, with a pair matching algorithm of polynomial complexity. Supervised classification with dimension reduction is tested on data bases of scrambled images, and for signals sampled on unknown irregular grids on a sphere.
🌉
Interdisciplinary Bridge
- Artificial Intelligence and Computer Science and Computer Vision and Mathematics & Optimization
📈
Trend Setter
- Graph Neural Networks
🐣
Hot Topic Early Bird
- graph neural network
Authors
Topics
Machine Learning > Core Methods > Representation Learning
Machine Learning > Learning Types > Unsupervised Learning
Deep Learning > Architectures > Neural Networks
Computer Science > Foundations > Algorithms
Machine Learning > Core Methods > Graph Neural Networks
Computer Vision > Processing > Image Processing
Mathematics & Optimization > Mathematics > Signal Processing
Machine Learning > Core Methods > Feature Extraction
Keywords
unsupervised learning
signal classification
graph classification
wavelet transform
deep learning
total variation
unsupervised classification
graph signal processing
haar scattering transform
signal descriptors
multiscale neighborhoods
haar wavelet
graph signal
scattering transform
graph neural network
haar scattering
invariant descriptor