Blind Construction of Optimal Nonlinear Recursive Predictors for
Discrete Sequences
C. Shalizi, and K. Shalizi. (2004)cite arxiv:cs/0406011Comment: 8 pages, 4 figures.
Abstract
We present a new method for nonlinear prediction of discrete random sequences
under minimal structural assumptions. We give a mathematical construction for
optimal predictors of such processes, in the form of hidden Markov models. We
then describe an algorithm, CSSR (Causal-State Splitting Reconstruction), which
approximates the ideal predictor from data. We discuss the reliability of CSSR,
its data requirements, and its performance in simulations. Finally, we compare
our approach to existing methods using variable-length Markov models and
cross-validated hidden Markov models, and show theoretically and experimentally
that our method delivers results superior to the former and at least comparable
to the latter.
Description
Blind Construction of Optimal Nonlinear Recursive Predictors for Discrete Sequences
%0 Generic
%1 shalizi2004blind
%A Shalizi, Cosma Rohilla
%A Shalizi, Kristina Lisa
%D 2004
%K hmm machinelearning
%T Blind Construction of Optimal Nonlinear Recursive Predictors for
Discrete Sequences
%U http://arxiv.org/abs/cs/0406011
%X We present a new method for nonlinear prediction of discrete random sequences
under minimal structural assumptions. We give a mathematical construction for
optimal predictors of such processes, in the form of hidden Markov models. We
then describe an algorithm, CSSR (Causal-State Splitting Reconstruction), which
approximates the ideal predictor from data. We discuss the reliability of CSSR,
its data requirements, and its performance in simulations. Finally, we compare
our approach to existing methods using variable-length Markov models and
cross-validated hidden Markov models, and show theoretically and experimentally
that our method delivers results superior to the former and at least comparable
to the latter.
@misc{shalizi2004blind,
abstract = {We present a new method for nonlinear prediction of discrete random sequences
under minimal structural assumptions. We give a mathematical construction for
optimal predictors of such processes, in the form of hidden Markov models. We
then describe an algorithm, CSSR (Causal-State Splitting Reconstruction), which
approximates the ideal predictor from data. We discuss the reliability of CSSR,
its data requirements, and its performance in simulations. Finally, we compare
our approach to existing methods using variable-length Markov models and
cross-validated hidden Markov models, and show theoretically and experimentally
that our method delivers results superior to the former and at least comparable
to the latter.},
added-at = {2019-03-30T19:21:51.000+0100},
author = {Shalizi, Cosma Rohilla and Shalizi, Kristina Lisa},
biburl = {https://www.bibsonomy.org/bibtex/2157dd67ad7897787ea496624eaaff97b/elfking},
description = {Blind Construction of Optimal Nonlinear Recursive Predictors for Discrete Sequences},
interhash = {59719cdaab403c3c0c04826ae64e45f3},
intrahash = {157dd67ad7897787ea496624eaaff97b},
keywords = {hmm machinelearning},
note = {cite arxiv:cs/0406011Comment: 8 pages, 4 figures},
timestamp = {2019-03-30T19:21:51.000+0100},
title = {Blind Construction of Optimal Nonlinear Recursive Predictors for
Discrete Sequences},
url = {http://arxiv.org/abs/cs/0406011},
year = 2004
}