VegML

Non-neural Machine Learning library implementation of Relational Probability Memory (RPM) supporting general inference and probabilistic information recall. VegML is a Vector Addressed Accumulated Memory that is utilized by RPMs to map relationships and retrieve relationship probabilities.

Vector Addressed Memory (VAM)

VegML can be used as a simple VAM or as a Vector Addressed Accumulated Memory (VAAM), where additions to the memory to be counted allowing frequency and probability to be assessed for Vectors and Vector-value pairs and sets.

VegML library

VegML library provides learning and inference that can easily be integrated into existing applications. VegML can be trained with datasets or can be programmatically trained with rules or direct data from you application. This versatility exists in retrieving predictions as well and allows code to be intermixed inside the decisioning process or managed in your application.

The examples provided in the code cover Natural Language Processing, Motion Control and Robot control with Webots. Additional examples are coming to show use in more general AI applications.

Documentation

Documentation is available in the Source Repository for VegML or follow the links to view the introductory PDF and research paper.

Additional information on Relational Probability Memory (RPM) can be found here.

Tutorial

Coming Soon…

Downloads

Coming Soon to our maven repo

Source Code

Coming Soon in github.com