RISO is software that offers an implementation of distributed, heterogeneous belief networks, allowing for efficient and collaborative learning and reasoning across multiple devices or systems.
RISO is a software that implements distributed, heterogeneous belief networks using Java. A belief network is a probability model defined on an acyclic directed graph, where distributed means nodes can be on different hosts, and heterogeneous means allowing different conditional distributions.
The software also involves multidimensional integrations, and if a partial result cannot be calculated from a catalog of special cases, RISO computes an approximate result by numerical integration. Messages are passed from one node to another as messages and via RMI if nodes live on different hosts.
The software comes with many example belief networks and lengthy documents to help users get started. In the latest "Jellyfish" release, minor bug fixes and functional enhancements have been implemented, including a Python module that can import RISO belief networks into Jython, which offers a flexible and elegant user interface.
Additionally, the compile.sh script now uses JavaDeps to generate a makefile and allows for recompiling one file at a time with javac, making it easier to use. The latest version also includes Poisson, exponential, and binomial distributions and initial support for considering a belief network as a multidimensional distribution. Finally, there is an initial version of a Gibbs sampler algorithm available.
The software also involves multidimensional integrations, and if a partial result cannot be calculated from a catalog of special cases, RISO computes an approximate result by numerical integration. Messages are passed from one node to another as messages and via RMI if nodes live on different hosts.
The software comes with many example belief networks and lengthy documents to help users get started. In the latest "Jellyfish" release, minor bug fixes and functional enhancements have been implemented, including a Python module that can import RISO belief networks into Jython, which offers a flexible and elegant user interface.
Additionally, the compile.sh script now uses JavaDeps to generate a makefile and allows for recompiling one file at a time with javac, making it easier to use. The latest version also includes Poisson, exponential, and binomial distributions and initial support for considering a belief network as a multidimensional distribution. Finally, there is an initial version of a Gibbs sampler algorithm available.
What's New
Version Jellyfish: N/A