Awesome

EmbeddedAI

EmbededAI is a 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. This library can dramatically simplify implementation of what otherwise would be complicated custom programming of hard to implement and maintain logic. With Bayesian approach the results are consistent with easily visualized model.

Methodology

1. Build Model
2. Apply sample/measurement
3. Query model for inference of non-directly observed variables
4. Query model for decisions based on maximum payoff/lowes loss

Applications

• Based on observations determine probabilities of any individual cause or combination of causes
• Find the most likely explanations of any given observation (Minimum Probability of Error MPE)
• Find most likely values of certain causes given observation (Maximum Aposteriori Probability MAP)
• Find optimal sequence of decisions (actions) given costs (payoffs) and observations performed prior to any action.

Doxygen pages of Library Components can be viwed here

API

Project Wiki contains brief API information. See doxygen pages here

Examples

Several test applications providing usage samples are available in ./tests directory and described in doxygen. Project wiki contains detailed walthrough of sample implementations.