ON THE GENERATIVE NATURE OF PREDICTION
Authors:
- 1 April 2009
- journal article
- research article
- Published by World Scientific Pub Co Pte Ltd in Advances in Complex Systems
- Vol. 12 (02) , 169-194
Abstract
Given an observed stochastic process, computational mechanics provides an explicit and efficient method of constructing a minimal hidden Markov model within the class of maximally predictive models. Here, the corresponding so-called ε-machine encodes the mechanisms of prediction. We propose an alternative notion of predictive models in terms of a hidden Markov model capable of generating the underlying stochastic process. A comparison of these two notions of prediction reveals that our approach is less restrictive and thereby allows for predictive models that are more concise than the ε-machine.Keywords
This publication has 11 references indexed in Scilit:
- Asymptotic Mean Stationarity of Sources With Finite Evolution DimensionIEEE Transactions on Information Theory, 2007
- Reductions of Hidden Information SourcesJournal of Statistical Physics, 2005
- Predictability, Complexity, and LearningNeural Computation, 2001
- Computational Mechanics: Pattern and Prediction, Structure and SimplicityJournal of Statistical Physics, 2001
- Observable Operator Models for Discrete Stochastic Time SeriesNeural Computation, 2000
- Theorie der stochastischen Automaten1995
- The calculi of emergence: computation, dynamics and inductionPhysica D: Nonlinear Phenomena, 1994
- Inferring statistical complexityPhysical Review Letters, 1989
- Toward a quantitative theory of self-generated complexityInternational Journal of Theoretical Physics, 1986
- On Stochastic Processes Derived From Markov ChainsThe Annals of Mathematical Statistics, 1965