IT-701 Soft Computing - Jun 2025 Question Paper

Under each question, How to answer lists the points an answer should cover, outlined by CampusPrep from the syllabus. Check the details against your textbook.

1. 7 Marks
   aDiscuss the architecture and training algorithm of multilayer feed-forward network.
   How to answer

   • Layers: input, one or more hidden, output with weighted interconnections
   • Forward pass computation with activation function
   • Backpropagation training steps: error calculation and weight update
   • Stopping criteria and generalization
2. 7 Marks
   bWith the help of a diagram, discuss the architecture of single layer feed-forward network.
   How to answer

   • Input layer directly connected to output layer with weights
   • Net input and activation / threshold operation
   • Neat labelled diagram of single layer network
   • Learning / limitation: only linearly separable patterns
3. 7 Marks
   aExplain architecture and training algorithm of ADALINE.
   How to answer

   • ADALINE architecture: inputs, weights, bias, linear combiner
   • LMS / Delta rule training steps and weight update formula
   • Diagram of ADALINE unit
   • Comparison with perceptron
4. 7 Marks
   bDiscuss various phases of back propagation algorithm.
   How to answer

   • Feed-forward phase: compute hidden and output activations
   • Backpropagation of error: output and hidden deltas
   • Weight and bias updation with learning rate
   • Iteration until error minimization
5. 7 Marks
   aDescribe the various processes of the Self-organizing Map algorithm.
   How to answer

   • Competition, cooperation and adaptation processes
   • Best matching unit selection and neighbourhood function
   • Weight update and learning rate decay
   • Topology preservation and convergence
6. 7 Marks
   bExplain various properties and operations of fuzzy set.
   How to answer

   • Definition of fuzzy set and membership function
   • Properties: commutativity, associativity, distributivity, De Morgan's laws
   • Operations: union, intersection, complement, difference
   • Example with membership values
7. 7 Marks
   aWhat do you mean by fuzzy composition? Discuss its types.
   How to answer

   • Definition of fuzzy relation composition
   • Max-min and max-product composition formulas
   • Types with example matrices
   • Applications in fuzzy inference
8. 7 Marks
   bExplain various methods of defuzzification.
   How to answer

   • Need for defuzzification from fuzzy to crisp output
   • Methods: centroid, weighted average, max-membership, mean-max, centre of sums
   • Formulas for each method
   • Comparison / suitability with example
9. 7 Marks
   aDerive Bayes rule. Discuss fuzzy Bayesian decision making.
   How to answer

   • Derivation from conditional probability: $P(A|B)=P(B|A)P(A)/P(B)$
   • Steps of Bayes theorem proof
   • Fuzzy extension with membership / possibility
   • Fuzzy Bayesian decision making procedure
10. 7 Marks
    bWhat is Classification? Discuss any one classification algorithm.
    How to answer

    • Definition of classification in pattern recognition / data mining
    • Steps: training and testing phases
    • One algorithm e.g. k-NN / decision tree / Bayesian classifier
    • Example application
11. 7 Marks
    aExplain different encoding schemes used in Genetic algorithm.
    How to answer

    • Role of encoding in genetic algorithm
    • Binary, octal/hex, permutation, value, tree encoding schemes
    • Example chromosome for each scheme
    • Advantages / suitability
12. 7 Marks
    bDiscuss the application areas of Neuro-fuzzy hybrid systems.
    How to answer

    • Concept of neuro-fuzzy hybridization
    • Application areas: control, pattern recognition, forecasting, medical diagnosis
    • Examples of ANFIS applications
    • Advantages of hybrid system
13. 7 Marks
    aWhat is the McCulloch-Pitts perceptron model? Does McCulloch-Pitts have ability of learning.
    How to answer

    • McCulloch-Pitts neuron model structure and threshold logic
    • Mathematical model with binary inputs and weights
    • Fixed weights - no learning ability explanation
    • Limitations and historical significance
14. 7 Marks
    bWhat is the RBFN method? Also discuss the back propagation network.
    How to answer

    • RBFN architecture: input, hidden radial basis, output layers
    • Gaussian basis function and training
    • Backpropagation network architecture and learning
    • Comparison of RBFN and backpropagation
15. 7 Marks
    aExplain the Adeline and Madeline briefly?
    How to answer

    • ADALINE: single adaptive linear element with LMS rule
    • MADALINE: multiple ADALINEs, hidden and output layers
    • Training rule for MADALINE (MRI, MRII)
    • Applications
16. 7 Marks
    bWhat are the advantages of ART network? Discuss about the Gain control in ART network.
    How to answer

    • ART network concept: stability-plasticity, vigilance
    • Advantages: online learning, stable, handles new patterns
    • Gain control mechanism: G1, G2 signals
    • Role in resonance and reset