Project Duality
Project Duality is an AI framework intended primarily for personally learning the fundementals of machine learning. Duality will run on an instance based system. This means that the entirety of an AI's knowledge can be saved to a file and ported to other machines. There will be two major file types used by Duality. You can find information on the formats and what their purpose is on the format information sheet.
Primary goals
These are the primary goals that I have set for the development of Project Duality. All of the goals listed in this category must be met before any public release of Duality will be offered.
-
Duality must provide a modular AI framework that is easy to work with in code.
-
The framework must be specifically designed for an environment of tinkering or any other method of testing.
-
The framework must also include the capability to learn from virtually any source of information.
-
The framework must offer a method of sending information to the learning system that can be modified by the user, but is otherwise automated in such a way that the user only has to define basic handling procedure.
-
Methods of monitoring and fine tuning the "neuron" connections, where viewing the score and altering the weight are straightforward processes.
-
-
Secondary goals
These are the secondary goals set for the development of Project Duality. These goals are not required for a successful design, but have great importance in the quality of the final product.
-
Duality should be able to run on hardware that may not be viewed as capable of machine learning.
- And, if applicable, potentially be run on portable devices.
-
Maintain an energy efficient model that will not introduce extraneous workload on hardware.