Home

Welcome to Embedded AI

EmbededAI is library that provides elements of AI to C++ applications. Due to small footprint it is usable even for small embedded systems. Initial set of algorithms in this library is based on Bayesian approach to AI.

Embedded AI 1.0

Initial realease

Features

• Encode Bayesian model of the system into software constructs. This include adding domain variables (states of the system), causal dependencies between variables and probabilities

• Use Bayesian reasoning to infer probabilities of states of unobserved variables based on partial knowledge of values of observed variables

• Find most likely explanation of observations for full set or subset of domain variables

• Add Software controlled actions and costs/payoffs associated with observed, unobserved variables and actions to them model

• Generate optimal actions to achieve most favorable state of Bayesian model based on partial knowledge of system state

The above algorithms are enhanced with optimizations:

• Node pruning

• Edge pruning

• Optimize sequence of variable elimination