AI Learning System

From Hive13 Wiki
Jump to navigation Jump to search

Overview

This is actually a series of several projects. This overall project is meant for data recording and for experimenting in artificial intelligence and learning systems. This can be used for the space but is also meant for fun and educational use only.

The AI Learning System is broken down into smaller sub categories.

Data Collection

We will need a database that can house all the data we can collect. For the most part we want the collected data to be similar to how a human would see it. Do not include a bunch of meta data. We will use other tables to implement meta information.

References

(TODO Links for data mining)

Learning Model

There are several different methods that an AI can "learn". One is simply by teaching it or training it the results. This can be done explicitly with constraints or programming. Preferably this can be done with a more "active learning" model. The more interesting and fun way for a computer to learn things and be "taught" information is through games and observation.

References

Learning Games

These are similar to googles image tagging game or other ones done by different universities (See references). Basically the AI selects from a pool of data. Some data it is fairly certain it knows the answer to and other data it has no idea. Then depending on the goal of the game, it will play with you and hopefully learn more information about the selected data. This information is categorized into meta data and stored in a database.

References

Goals

This is where the AI can be invoked to use what it's learned against the data it's collected to provide some kind of useful output.