Classification is one of the important topics in AI. All agents will perform this type in one way or the other.
What is Classification?
It is like extracting a set of concepts into a sequential set of classes.
###How can it be done?
There are three major type of concept classification as below
- Axiomatic concepts
- Prototype concepts
- Exemplar concepts
These are the set of concepts which have proper definition or formal set of conditions. These are easily programmable via computer agents. Eg Definition of a circle
These are like base concepts which will be overrided to its own values for other set of concepts. This is little difficult to program via computer agents Eg Chair Definition
These are the general abstractions of the particular concepts. Very Very difficult to program via computer agents Eg defintion of beauty could be a painting, flower, natural scenary or beautiful girl
###Top Down Approach
In classification a bird we use top down approach where we know that birds belongs to vertebrate and in vertrabate we have more types and birds is one among them . With this background knowledge we are able to classify the birds
###Bottom Up Approach
Bottom up approach is a method in which we try to learn the concept of the bottom tree values based on which we abstract to base root values to get the concept.
One typical example of cognitive connection of classification to humans is the way how doctors classify the symptoms to a disease and based on that medication is provided
Another example is how developers approach the bugs in the programs. They classify under different type and provide fix for the bugs based on background knowledge.