Fuzzy logic

Question 1

Which type of AI is about precision? and which one about significance?

Choose between symbolic and sub-symbolic.

Answer 1

Precision: sub-symbolic
Significance: symbolic

Question 2

What is fuzzification?

Answer 2

The process of mapping a numeric value to a fuzzy set through a membership value in the [0,1] interval.

Question 3

What is the triangular membership function?

Answer 3

Can be defined by three points that form a triangle (a,b and c). Only one point has value 1.
Formula: (x - a)/(b - a)

Question 4

What is the trapezoidal membership function?

Answer 4

Can be defined by four points (a,b,c,d), which form a trapezium.
All points in [b,c] reach max membership (1).

Question 5

What is the difference between fuzzy logic and probabilities?

Answer 5

Fuzzy logic relies on degrees of truth as a mathematical model of vagueness and imprecision while probabilities are a mathematical model of ignorance.

Question 6

What is convergence?

Answer 6

We say that an FCM-based model has converged (to a fixed-point attractor) if, after performing a large enough number of iterations, the neurons continue to produce the same activation values over and over again.

Question 7

What is a unique fixed-point attractor?

Answer 7

If this FCM model converges to a unique fixed-point attractor, the model will produce the same output regardless of the inputs.

Question 8

What is instantation?

Answer 8

Assigning a value to a symbolic variable

Question 9

How can constraint problems be solved?

Answer 9

By using backtracking to implement the generate-and-test procedure

Question 10

What are the two golden rules of the generate-and-test method regarding optimization?

Answer 10

1. We should avoid guessing variables that can be guessed by other variables
2. We should avoid placing separate guesses between generation and testing of values.

Question 11

What does a counterfactual explanation do?

Answer 11

It speculates an alternative outcome to past events.