EC-802-A AI and Signal Processing - 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
   aWhat is Artificial Intelligence (AI)? How is AI used in real-world applications.
   How to answer

   • Define Artificial Intelligence as simulation of human intelligence by machines
   • List core capabilities: reasoning, learning, perception, NLP, robotics
   • Give real-world applications: healthcare, finance, autonomous vehicles, smart assistants, signal processing
   • Mention benefits and impact with examples
2. 7 Marks
   bHow does AI differ from Machine Learning (ML) and Deep Learning (DL)?
   How to answer

   • Define AI as broad field, ML as subset learning from data, DL as subset using deep neural networks
   • Compare on basis: scope, data need, feature engineering, compute, interpretability
   • Give examples for each level
   • Summarize hierarchical relationship AI > ML > DL
3. 14 Marks
   What are the trade-offs between expressiveness and computational efficiency in knowledge representation?
   How to answer

   • Define knowledge representation and its goals
   • Explain expressiveness: ability to represent complex facts, uncertainty, logic
   • Explain computational efficiency: reasoning speed, tractability
   • Discuss trade-off with examples: FOL vs propositional logic vs semantic nets/frames
4. 7 Marks
   aWhat is knowledge in AI, and why is it important?
   How to answer

   • Define knowledge in AI: facts, rules, heuristics acquired and represented
   • Explain types: declarative, procedural, meta-knowledge
   • Explain importance for reasoning, decision making, expert behaviour
   • Give example of knowledge use in expert system
5. 7 Marks
   bWhat are the key components of a knowledge-based system? What are the advantages and limitations of knowledge-based systems?
   How to answer

   • List components: knowledge base, inference engine, user interface, explanation facility, knowledge acquisition
   • Explain advantages: consistency, preservation of expertise, availability
   • Explain limitations: knowledge acquisition bottleneck, brittleness, maintenance cost
   • Conclude with application suitability
6. 7 Marks
   aHow are Forward and Backward Chaining used in Expert Systems?
   How to answer

   • Define forward chaining as data-driven from facts to conclusions
   • Define backward chaining as goal-driven from hypothesis to facts
   • Explain working in expert systems with rule example
   • Compare use cases and inference control
7. 7 Marks
   bWhat are the challenges in generalization and overfitting?
   How to answer

   • Define generalization vs overfitting and bias-variance trade-off
   • List challenges: limited data, noise, model complexity, distribution shift
   • Explain consequences on unseen data performance
   • Mention mitigation: regularization, cross-validation, more data
8. 7 Marks
   aWhat is a Decision Tree, and how does it help in learning? and how does the ID3 algorithm construct decision trees?
   How to answer

   • Define decision tree: nodes, branches, leaves for classification/regression
   • Explain how it aids learning via recursive partitioning and interpretability
   • Outline ID3 steps: entropy, information gain, best attribute split, recursion
   • Mention limitations and need for pruning
9. 7 Marks
   bWhat is overfitting in decision trees, and how can it be prevented?
   How to answer

   • Define overfitting in decision trees: deep tree memorizing training data
   • Explain causes: noisy data, too many splits, small leaf samples
   • List prevention: pre-pruning, post-pruning, max depth, min samples, cross-validation
   • Mention ensemble alternative like random forest
10. 14 Marks
    What is the difference between supervised, unsupervised, and reinforcement learning?
    How to answer

    • Define supervised learning with labeled data and examples
    • Define unsupervised learning with unlabeled data for clustering/dimensionality reduction
    • Define reinforcement learning with agent, reward, environment
    • Compare on data, feedback, goal and applications in table
11. 7 Marks
    aWhat is the difference between upsampling and downsampling?
    How to answer

    • Define upsampling as increasing sampling rate by interpolation
    • Define downsampling as decreasing rate by decimation
    • Compare process, anti-imaging vs anti-aliasing filtering
    • Give applications in multirate signal processing
12. 7 Marks
    bWhat are the challenges in changing the sampling rate of a signal?
    How to answer

    • Explain need for anti-aliasing and anti-imaging filters
    • Discuss issues: aliasing, imaging, phase distortion, computational load
    • Mention fractional rate change and filter design complexity
    • Mention preservation of signal integrity
13. 7 Marks
    aHow are Finite Impulse Response (FIR) filters used in polyphase decomposition?
    How to answer

    • Define FIR filter and polyphase decomposition concept
    • Explain splitting FIR into M polyphase sub-filters for decimator/interpolator
    • Show computational saving by filtering at lower rate
    • Mention noble identities and efficient structures
14. 7 Marks
    bWhat is polyphase decomposition, and how does it work?
    How to answer

    • Define polyphase decomposition as splitting sequence/filter into phases
    • Give mathematical form $H(z) = \sum_{k=0}^{M-1} z^{-k}E_k(z^M)$
    • Explain working for decimation and interpolation structures
    • Highlight efficiency and implementation benefits