|Intelligent Systems And Their Societies||Walter Fritz|
The pattern finder is a rule, composed of several rules, which refers to computer code. The pattern finder is activated when the concept “sleep” exists in the internal situation. Each of the several rules of the pattern finder, is activated by an additional concept in the internal situation. Each rule cancels the existing additional concept and puts its own into the interior situation. This means that it has done its job and by this concept the brain selects the next rule.
The brain needs a minimum of 3 examples to find a pattern. Once it finds a pattern (a regularity), it creates new concepts and rules accordingly.
A rule of type “instruction” indicates what modification has to be done to the future situation of a rule.
Here are the rules
Make rule list
Takes the last rule from the chronological memory and makes a list of all rules of the general memory, that have some concept in their initial situation identical to a concept in the initial situation of the “last rule”. With this list it cycles in a loop through the rest of the rules of the pattern finder.
At the end of the cycle it eliminates the last rule of the chronological memory. Now it starts over again with the new “last rule”.
If there are several rules in this list, that have the same group of concept´s in the initial situation, then it creates a combined concept for this group and creates a new rule where the combined concept replaces the individual concepts.
If the future situation of a rule is identical to the initial situation of another rule, then it creates a new rule combining the two. The new rule has as initial situation, the initial situation of the first rule and as future situation the future situation of the second rule. The situation that is identical to both rules is stored in the intermediate situation.
If there are several rules, that differ only by one concept in the initial situation, then it creates an abstract concept with a content of all these concepts as concretes. Further it creates a new rule having in the initial situation the new abstract concept in place of the concrete concept. It does the same for the future situation.
Generalize the situation
If there are several rules, that have identical future situations, and have only some concept´s identical in the initial situation, then it creates a rule with only these concept´s in the initial situation.
Make "same concept" rules
If there are several rules, that have the same concept in the initial situation and the future situation, then it creates a rule of type ”instruction”. Also it creates a rule that has the label of this instruction rule in place of the “same” concept´s in the initial and the future situation. The rule of type “instruction” indicates that whatever label is found at this place in the initial situation, has to be put into the future situation replacing the label of the instruction rule.
Make repeated concept rules
If there are several rules, that have the same concept in the initial situation and all have a different concept in the future situation (but common to all of these rules), then it creates a rule of type “instruction”. Also it creates a rule that has the label of this instruction rule in place of the “same concept” in the initial situation and the future situation.
For theoretical considerations on patterns, see here
This version of the pattern finder was developed for a program with text and sound input and output. It will have to be enlarged for vision.
Also, at present, a limited amount of rules that look for patterns are installed. In the future all possible types of patterns between rules should be defined and rules to detect them included, to make sure that any possible pattern can be learned.
If you liked this page, please do me a favor: Tell your friends about it.
The author appreciates comments and suggestions.
For continuous reading, like a book - continue at
Jump to the
E-Book Contents /
/ Robot Brain
/ top of this document
Copyright © New Horizons Press