Q1.(10)

Write a state space solution for following problem. Write clearly; What are Initial State, Final State, Rules, control strategy for this problem. Write generalized set of rules for the same. Can you indicate a heuristic for solving the problem faster?

A hungry monkey finds himself in a room where a bunch of bananas is hanging from the ceiling. The monkey, can not reach the bananas directly. The ceiling is the just the right height that monkey standing on a chair, weaving stick can knock the bananas down. The monkey knows how to move around, carry other things around, reach for the bananas, and wave a stick in the air. Get the rules to solve the problem and help the monkey out.

Q2. Explain following terms in brief (Any ten).(10)

1.physical symbol system hypothesis

2.A non monotonic production system

3.Frame problem

4.“Annealing Schedule” in simulated annealing

5.Closed nodes in case of A* or AO* algorithms

6.Computable Predicate

7.parallel relaxation

8.Generalization

9.Competitive Learning

10.Extensional Programming

11.Ultra Fuzzy Sets

12.Conditional proof justification

Q3. Write answers to following in brief (Any 5)                                       (10)

1.Why changing a single fact in NMRS forces lot of revision?

2.Why it is important to check if the problem is decomposable or not?

3.What is the difference in the knowledge that is needed in recognizing a solution and knowledge needed in constraining a search?

4.What are the techniques possible to use to make normal generate and test useful for real world applications?

5.Why ‘Best First Search’ has two values g and h’?

6.How traveling salesman problem is solved using Hopfield networks?

7.Why predicate logic fails to represent some of the facts.

Q4. Write whether true or false. Write reasons for the same. (Any 10)          (10)

1.P(Bi/E) is same as P(E/Bi)

2.Perceptron convergence theorem says that we must get a decision surface for the problem if one exists.

3.Our brain employ massively parallel algorithm to solve visionary problems.

4.In competitive learning the weights from all input units are rationed to 1?

5.Adjusting the shape of an activation function can have an effect on network’s susceptibility to local minimum

6.Longer paths are better at times in AO*

7.Initially p’ value is more and later on reduced to 0 in simulated annealing

8.Heuristics are used if they do not guarantee the best solution.

9.Fuzzy Logic is not fuzzy

10.Truth maintenance system is considered ‘passive’

11.CF is not a probabilistic measure

Q5. Solve following                                                                          (10)

1.Represent following facts in predicate logic and prove ‘ john likes peanuts’

a. John likes all kinds of food

b.apple are food

c.chicken is food

d.Anything anyone likes and isn’t killed by is food.

e.Bill eats peanuts and still alive.

2.Assume following rules are given for an auto pilot

a.More height then normal, push the throttle a bit

b.More speed then normal, push the throttle a bit

c.Running into clouds, have more height.

d.To have more height, pull the throttle a bit

Show how autopilot takes an action when running into clouds.