1) Design the delta learning rule for the multi-layer perceptron (utilizing the error back-propagation), that updates the weight wji connecting the neuron i to neuron j. suppose that the activation functions within the network are continuous. Consider cases of:
a) j is an output neuron
b) j is a neuron in a hidden layer
2) Describe the architecture of ANFIS model with the help of a neat and well labelled diagram.
3) Describe the operation of the Fuzzy Logic Controller by giving the simple example.
4) Design a genetic algorithm for solving any problem of your choice, for example: Travelling Sales Person problem. Explain precisely the bit-string encoding and the set of crossover operators. Then, provided a data set, describe the procedure of utilizing the genetic algorithm to deduce the solution and also the stopping criterion.
5) Consider the strings and schemata of length 11. For following schemata determine the probability of surviving mutation if probability of the mutation is 0.001 at the single bit position. **100****10, 0**********1, 11***00***1, *1111*0000*. Re-determine the survival probabilities for the mutation probability Pm=0.1.
6) Explain what is wrong with the following given argument:
a) Men are widely distributed over the earth.
b) Socrates is the man.
c) Thus, Socrates is widely distributed over the earth.
How should facts be represented by these sentences be represented in the logic so that this problem does not occur?
7) Create the partitioned semantic net representation for following:
a) Every batsman hit a ball.
b) All batsmen like the wicket-keeper.
9) How can the development of an expert system be viewed as the core software engineering process? Describe with suitable example to support your discussion.
10) Describe the detailed life cycle model of the expert system.
11) Explain the attributes of the several components of the typical rule-based expert system.