Online learning of Riemannian hidden Markov models in homogeneous\n Hadamard spaces

2021/02/15 by Quinten Tupker, Tupker, Quinten, Salem Ben Saïd +3 · 1 citation
Computer Science · #Bayesian Methods and Mixture Models #Face and Expression Recognition #Topological and Geometric Data Analysis

paper · pdf · doi:10.48550/arxiv.2102.07771

Abstract

Hidden Markov models with observations in a Euclidean space play an important\nrole in signal and image processing. Previous work extending to models where\nobservations lie in Riemannian manifolds based on the Baum-Welch algorithm\nsuffered from high memory usage and slow speed. Here we present an algorithm\nthat is online, more accurate, and offers dramatic improvements in speed and\nefficiency.\n

Cited by

• Computing functions of A^-1B where A and B are Hermitian matrices

Related

• Hidden Markov chains and fields with observations in Riemannian\n manifolds
• On the Riemannian barycentre of a Markov chain
• Riemannian Denoising Diffusion Probabilistic Models
• DenseHMM: Learning Hidden Markov Models by Learning Dense Representations
• Implicit Riemannian Optimism with Applications to Min-Max Problems