About this previous year paper
Subject Code: BEE056
Subject Name: Neural Networks & Fuzzy System
Semester: V (Fifth Semester, B.Tech)
Maximum Marks: 70
Time Duration: 3 Hours
Exam Year: 2024–25
Paper Structure: Three Sections – A, B, and C
SECTION A – Short Answer Questions (2 × 7 = 14 Marks)
Attempt all questions. Each is conceptual and brief.
Explain the perceptron rule.
Describe unsupervised learning.
Write a short note on Madaline networks.
Write the factors affecting backpropagation training.
Explain product of two fuzzy sets with an example.
Write the fuzzy output variables for an air conditioner.
Write the factors affecting performance of fuzzy BP networks.
SECTION B – Medium-Length Questions (7 × 3 = 21 Marks)
Attempt any three of the following:
Develop the structure of the human brain with the help of a neuron diagram.
Explain single-layer and multi-layer perceptron networks.
Write short notes on:
(i) Fuzzy arithmetic
(ii) Fuzzy to crisp conversion
Explain membership function in fuzzy logic. What is inference in fuzzy logic?
Define the concept of a fuzzy neuron and explain how it differs from traditional and probabilistic models.
SECTION C – Long/Analytical Questions (7 × 5 = 35 Marks)
Attempt one part from each question.
Q3
a. With a neat sketch, explain activation function and its use in neuron models.
b. Write the expression for bipolar continuous and bipolar binary activation functions.
Q4
a. Discuss the Rosenblatt’s perceptron model.
b. Generate the output of AND logic function using McCulloch-Pitts neuron model.
Q5
a. Explain crisp set operations.
b. Compare and contrast classical logic and fuzzy logic.
Q6
a. Explain the fuzziness of a fuzzy set and define a fuzzy function.
b. Explain different methods of defuzzification.
Q7
a. Explain the inference process in fuzzy BP (backpropagation). How is fuzziness propagated through layers?
b. Discuss how L-R type fuzzy numbers are used in decision-making to address uncertainty.
Key Topics for Preparation
Perceptron models (Single and Multi-layer)
Backpropagation algorithm and training factors
Madaline and McCulloch-Pitts models
Fuzzy logic concepts: sets, membership functions, fuzzification, and defuzzification
Fuzzy arithmetic and fuzzy inference systems
Neural–fuzzy hybrid models
Application of L-R fuzzy numbers in decision-making
Study Tips
Focus on understanding neural model architectures and their mathematical functions.
Revise fuzzy logic operations (union, intersection, complement, product).
Practice drawing diagrams — especially neuron models and fuzzy inference systems.
Memorize key differences between classical, probabilistic, and fuzzy systems.
Go through example problems on defuzzification methods (centroid, mean of maxima, weighted average).