Recency-weighted Markovian inference
Machine Learning
2017-11-09 v1 Machine Learning
Abstract
We describe a Markov latent state space (MLSS) model, where the latent state distribution is a decaying mixture over multiple past states. We present a simple sampling algorithm that allows to approximate such high-order MLSS with fixed time and memory costs.
Keywords
Cite
@article{arxiv.1711.03038,
title = {Recency-weighted Markovian inference},
author = {Kristjan Kalm},
journal= {arXiv preprint arXiv:1711.03038},
year = {2017}
}