Certainty Factor. How does it help deal with uncertainty? Explain w.r.t rule based systems.

Rule based systems can be augmented to draw variably certain conclusions.

The basic idea is to add Certainty Factors to rules, and use these to calculate the measure of belief in some hypothesis.

Certainty Factors are related to conditional probability, but are not the same.

A Certainty Factor involves a measure which represents a level of confidence that some condition is true.

They consist of two components – A measure of belief (Step of opinion) and a measure of disbelief (Step of incredulity).

In step of opinion, given the hypothesis (H) and evidence (E), measure of belief is the extent to which the evidence supports the hypothesis.

In step of incredulity, given the disbelief hypothesis (H) and evidence (E), measure of disbelief is the extent to which the evidence supports the negative of hypothesis.
                 Certainty factor [Hypothesis, Extent]
                                           =
   Measure belief [Hypothesis, Extent] – Measure disbelief [Hypothesis, Extent]

CF in rule based systems:

A rule based system can propagate CFs via rules, to give CFs to conclusions.

For this, rules too may have CFs associated with them.

Ex.:
RULE [Is the petrol tank empty?]
If [the result of trying the starter] = “The car cranks normally” and [a petrol smell] = “Not present when trying the starter”
Then [the petrol tank] = “empty” @90.

This says that if we know that the 2 conditions in the premise hold, then this knowledge gives a degree of belief that [the petrol tank] has the value “empty”.

If no CF is given for a rule, it is assumed to be 100%.

No comments:

Post a Comment