Can Fuzzy learn or just deduce?


(0 comments)

When it comes to Machine Learning, people often think of Neural Network. Indeed, the power of the Neural Network is very strong. However, the application of Neural Network is limited to some fields that require a large number of samples.

Fuzzy can be applied to any problem that an expert offers a set of rules. Fuzzy not only deduced but also learned. It forms new rules that are not included in the rules set by the programmer. To illustrate this, let's look at the entire set of rules in the Sail solution:

rule max_temp.heat OR LITTLE min_temp.freeze => dangerous;
rule LITTLE LITTLE storm.violent OR BIT storm.strong OR wind.gale => dangerous;
rule (REALLY wind.strong OR REALLY storm.medium) AND
    EXTREMELY NOT (max_temp.heat OR LITTLE min_temp.freeze) => bad;
    
rule thunder.yes AND (wind.moderate OR storm.light) AND
    EXTREMELY NOT (max_temp.heat OR LITTLE min_temp.freeze) => bad;
    
rule SEEMED max_temp.mild AND SEEMED min_temp.mild AND storm.none AND
    (LITTLE LITTLE wind.calm OR BIT wind.breeze) AND
    thunder.yes AND fog.no => normal;    
    
rule REALLY fog.yes AND EXTREMELY NOT (max_temp.heat OR LITTLE min_temp.freeze) AND
    VERY VERY VERY NOT (LITTLE LITTLE storm.violent OR BIT storm.strong OR
    LITTLE wind.gale) => bad;
    
rule (max_temp.hot OR min_temp.cold) AND storm.none AND
    (LITTLE LITTLE wind.calm OR BIT wind.breeze) AND
    thunder.no AND fog.no => normal;
    
rule SEEMED max_temp.mild AND SEEMED min_temp.mild AND storm.light AND
    (LITTLE LITTLE wind.calm OR BIT wind.breeze) AND
    thunder.no AND fog.no => normal;

rule max_temp.mild AND min_temp.mild AND storm.none AND
    wind.moderate AND thunder.no AND fog.no => normal;

rule LITTLE LITTLE max_temp.mild AND LITTLE LITTLE min_temp.mild AND
    storm.none AND (LITTLE LITTLE wind.calm OR BIT wind.breeze) AND
    thunder.no AND fog.no => good;


There are 10 rules in which only the sixth rule assumes foggy (fog.yes) with the conclusion that bad weather (bad). But Fuzzy has learned a rule that is understood as follows:
"In any case, when the weather is foggy, the weather is worse."
Try a case:

The arguments read into the program are as follows, from left to right:

  • 44: The highest temperature of the day is 44°C
  • 36: The lowest temperature in the day is 36°C
  • 0: There is no storm
  • 10: Wind speed is 10km/h
  • 0: There is no thunder
  • 0: There is no fog

The result is Bad weather due to too hot. However, that is not the worst level of weather.
Now we only change the fog factor, becoming foggy

The result is Danger - the worst level of weather.

If only inference, Fuzzy will tend to conclude the weather is Bad like the sixth rule. But it learned and decided the result was Danger.

Currently unrated

Comments

There are currently no comments

New Comment

required

required (not published)

optional

required


What is 7 × 2?

required