Transport-based analysis, modeling, and learning from signal and data\n distributions

2016/09/15 by Soheil Kolouri, Serim Park, Kolouri, Soheil +7
Chemistry · Computer Science · Environmental Science · #AI in cancer detection #Bayesian Methods and Mixture Models #Computer Vision and Pattern Recognition (cs.CV) #Electrostatics and Colloid Interactions #FOS: Computer and information sciences #Hydrological Forecasting Using AI #Image Retrieval and Classification Techniques #Medical Image Segmentation Techniques

paper · pdf · doi:10.48550/arxiv.1609.04767

openalex publication_date 2016/09/15 · openalex created_date 2022/08/30 · openalex updated_date 2026/07/28

Abstract

Transport-based techniques for signal and data analysis have received\nincreased attention recently. Given their abilities to provide accurate\ngenerative models for signal intensities and other data distributions, they\nhave been used in a variety of applications including content-based retrieval,\ncancer detection, image super-resolution, and statistical machine learning, to\nname a few, and shown to produce state of the art in several applications.\nMoreover, the geometric characteristics of transport-related metrics have\ninspired new kinds of algorithms for interpreting the meaning of data\ndistributions. Here we provide an overview of the mathematical underpinnings of\nmass transport-related methods, including numerical implementation, as well as\na review, with demonstrations, of several applications.\n

Citations

• Wasserstein Training of Boltzmann Machines
• Visual Domain Adaptation: A survey of recent advances
• A computational fluid mechanics solution to the Monge-Kantorovich mass transfer problem
• Remarks on a Multivariate Transformation
• Mathematical Methods of Organizing and Planning Production
• Deep Learning in Neural Networks: An Overview
• The geometry of optimal transportation
• Polar factorization and monotone rearrangement of vector‐valued functions
• Reconstruction of the early Universe as a convex optimization problem
• The Earth Mover's Distance as a Metric for Image Retrieval
• Convex Color Image Segmentation with Optimal Transport Distances

Related

• Optimal Mass Transport: Signal processing and machine-learning applications
• Data representation with optimal transport
• Partitioning signal classes using transport transforms for data analysis and machine learning
• Transport-Based Pattern Theory: A Signal Transformation Approach
• Embedding Empirical Distributions for Computing Optimal Transport Maps