For us artificial learning involves the:
-
- reading and listening to information from any source;
- determining what is important;
- verifying it;
- categorizing and organizing it;
- storing it for different types of access; and
- applying it to different situations when requested.
Depending what that information is and when/where it is received, that may trigger one or more of our Artificial Assistants to act on that information. For example, a client may have requsted notification if a particular web page is changed, or a filing is made on a government web site. The Artificial Assistant would then notify the client of the event.
Summarizing, our focus is on learning how to do something and doing it when appropriate. Appropriate could mean when directed or when an artificial assistant based on previous history or guidance decides to take some action. For us the artificial learning process is divided into 3 parts:
-
- Identifying that there is an issue (e.g., environmental issues, human command, notification, …);
- Determining what are the possible solutions and actions that can be taken;
- Taking action to resolve and close the issue.
So our intelligent program needs to undertand the issue (learning part 1), determine what the potential solutions are (analysis part 1, issue attribute probing part 1, learning part 2 – determine possible solutions), and resolving the issue (learning part 3 – learn how to do something; analysis part 2 – determining whether the issue was resolved; learning part 4 – either reinforce the fact that this solution resolved the issue, or trigger learning part 5 – what are other alternatives). This process and its requisite components are evolving and growing over time and use.