Online learning of Riemannian hidden Markov models in homogeneous Hadamard spaces
This work addresses computational bottlenecks in signal and image processing for applications involving manifold-based data, representing an incremental improvement over existing methods.
The authors tackled the problem of high memory usage and slow speed in Riemannian hidden Markov models by developing an online algorithm, which resulted in improved accuracy and dramatic efficiency gains.
Hidden Markov models with observations in a Euclidean space play an important role in signal and image processing. Previous work extending to models where observations lie in Riemannian manifolds based on the Baum-Welch algorithm suffered from high memory usage and slow speed. Here we present an algorithm that is online, more accurate, and offers dramatic improvements in speed and efficiency.