PGM sampling is the core of our Bayesian Risk Systems. This module offers a fast, flexible and scalable computation of the conditional probability distributions over the variables of interest for any kind of node type and distributions.
Features
Multiple node types: Categorical, Discrete, Continuous, Mixture, Deterministic.
-
Exhaustive list of Discrete Distributions including: Binomial, Geometric, Hypergeometric, Logseries and Poisson.
Exhaustive list of Continuous Distributions including: Beta, Cauchy, Chisquare, Exponential, Gamma, Gumbel, Laplace, Logistic, Lognormal, Non Central Chisquare, Normal, Pareto, Power, Rayleigh, Student-t, Triangular, Uniform, Vonmises, Wald and Weibull.
Expression Parser for custom formulas combining operations on nodes.
Support for hybrid networks.
Export sampled graph values.
Multi step simulation for hybrid networks.
References
- Koller D, Friedman N. Probabilistic graphical models: principles and techniques. MIT press; 2009.
- Sucar LE. Probabilistic graphical models. Advances in Computer Vision and Pattern Recognition. London: Springer London. doi. 2015;10(978):1.
- Bessière P, Mazer E, Ahuactzin JM, Mekhnacha K. Bayesian programming. CRC press; 2013 Dec 20.
