| NEURAL NETWORKS AND FUZZY LOGIC |
| PAPER NO. 1 |
 |
 |
 |
|
| PUT ON: May ,2002 |
|
EC - 44 NEURAL NETWORKS AND FUZZY LOGIC (B.Tech 8th Semester, 2052) Time : 3 Hours Maximum Marks : 60 NOTE:- This paper consist of Three Sections. Section A is compulsory. Do any Four questions from Section B and any two questions from Section C Section-A Marks : 20 1(a) Compare and contrast a biological neuron and an artificial neuron. (b) Explain the role of bias in activation function. (c) What is the significance of initial weighta and learning rate in training of ANN ? (d) Draw Mc-Culloch - Pitts neuron model. (e) Justify the name "error backpropagation" of EBP. (f) In which networks "winner" node concept is used ? (g) Does fuzzy logic mean fuzzy input or fuzzy logic ? (h) Name few consumer products in which fuzzy electronics is used. (i) If an item has partial membership in several sets, would all membership values add up to unity ? (j) Where are Radial Basis functions networks preferred in ANN ? Section-B Marks:5 Each 2. Using perceptron model of ANN design an OR gate. 3. Write a note on CMAC Nework. 4. Write a note on Counter Propagation Network. 5. Write a note on engineering applications of Fuzzy control. 6. Compare different defuzzification methods. Section-C Marks : 10 Each 7. Design a fuzzy control system for a break system of a car. 8. Describe with the help of an example how ANNs and what architecture can be used for the following applications : (a) Optimization (b) Mul 9. What are RBF-nets and what are the node function of these nets ? Write the algorithm for RBF network for function approximation. |