Scribd
Upload
159 views2 pages

Final Exam Artificial

The document is an artificial intelligence final exam containing 6 questions that assess various AI topics. Question 1 covers classical vs fuzzy rules, hedges, error gradient, and backpropagation drawbacks. Question 2 asks for a Hebbian learning flowchart. Question 3 involves applying competitive learning to weight vectors. Question 4 deals with a fuzzy control system for error and change in error. Question 5 compares a probe to Hopfield network memories. Question 6 uses a genetic algorithm to select a neural network architecture for an AND operation.

Uploaded by

ZaidB
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
159 views2 pages

Final Exam Artificial

The document is an artificial intelligence final exam containing 6 questions that assess various AI topics. Question 1 covers classical vs fuzzy rules, hedges, error gradient, and backpropagation drawbacks. Question 2 asks for a Hebbian learning flowchart. Question 3 involves applying competitive learning to weight vectors. Question 4 deals with a fuzzy control system for error and change in error. Question 5 compares a probe to Hopfield network memories. Question 6 uses a genetic algorithm to select a neural network architecture for an AND operation.

Uploaded by

ZaidB
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Electrical & Computer Engineering Department

Final Exam
Course: Artificial Intelligence Course Code:
Session 2014/2015 Date : 4/3/2015
Lecturer: Lokman H Hassan Duration : 180 min

Q1: (18 marks)


1. Explain the difference between classical and fuzzy rules? Give an example
2. Explain with examples the hedges performing operations of concentration.
3. What is the error gradient?
4. Explain the main drawbacks of the back-propagation learning.
5. The possibility of GA’s trapped in local optima can be reduced. How?
6. What is crossover in GA? Explain two of crossover options that supported by MATLAB.

Q2: (6 marks)
Draw a flowchart showing the main steps of Hebbian learning algorithm. Define all parameters. What is
the meaning of forgetting factor?

Q3: (8 marks)
Consider a two-dimensional input vector as
𝑝 = [1 −1]
and the initial weight vectors as
−0.543 −0.969 0.997 0.456
𝑤 = ;𝑤 = ;𝑤 = ;𝑤 =
0.840 −0.249 0.094 0.954
Apply the competitive learning network training algorithm to obtain the updating weight vectors. Assume
that the learning rate parameter is equal to 0.5. Train the network for only one iteration.

Q4: (16 marks)


A fuzzy control system use inputs of error (𝑒) and change in error (∆𝑒) to control an output variable 𝑢.
The fuzzy algorithm and characteristics of the fuzzy membership functions are given below. Determine
the output 𝑢 for e = -1.65% and ∆𝑒 = +4.95%/min using a zero-order Sugeno inference method. The
output 𝑢 is equal to 𝐾 × 𝑒. Use the Weighted average method to defuzzify the answer. Use product AND
method to evaluate the conjunction of the rule antecedents. Show a schematic representation of the fuzzy
inference. The fuzzy rules are in the form:

If e is A1k AND e is A2 k Then K is C k , k  1,2,, n

where A1k , A2 k and C k are fuzzy sets.


Fuzzy logic rules for K
e
e NB NM NS Z PS PM PB
NB M B B B B B M
NM S M B B B M S
NS S M M B M M S
Z S S S S S S S
PS S M M B M M S
PM S M B B B M S
PB M B B B B B M

Q5: (7 marks)
Store two fundamental memories (−1 1 1) and (−1 1 −1) in the three-neuron Hopfield network.
Compare the probe (1 −1 1) with the fundamental memories.

Q6: (15 marks)


Using a three layer back-propagation network to perform AND operation. Use a genetic search algorithm
to select the suitable neural network architecture. Apply the genetic operators (i.e. selection, crossover and
mutation) for only one iteration. Train the network for one epoch using back-propagation algorithm. Use
Roulette wheel function and Single point function respectively to perform the selection and crossover.
Assume that mutation takes place in a digit of a chromosome.

Good Luck

You might also like