Analysis of Probabilistic Graphocal Model Based on Markov Random Field and Conditional Probability
Download as PDF
DOI: 10.38007/Proceedings.0000027
Author(s)
Genwang Zhou, Jian Zhou and Saihan Wang
Corresponding Author
Genwang Zhou
Abstract
This paper discusses the Probabilistic Graphocal Model (PGM) based on Markov random field and conditional probability. Through extensive verifications, it can be found that a random variable of the Markov random field is only related to its adjacent random variables, and is independent of random variables that are not adjacent. Furthermore, we analyze the potential function of Markov random field, Markov network connections are often expressed as a log-linear model. Finally, we also focus how the model to represent the uncertainty in the real world and the relationship between the various quantities. It is clear that the joint probability of the transition probability and the performance probability. The statistical probability is the conditional probability. Because it is only normalized locally, it is easy to fall into the local optimum.
Keywords
Probabilistic Graphocal Model; Markov Random Field; Experiment Analysis