AD-502 Machine Learning - Nov 2022 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
   aBriefly explain the types of Machine Learnings.
   How to answer

   • Define machine learning
   • List three main types: supervised, unsupervised, reinforcement (with semi-supervised if expected)
   • Brief working and example of each
   • Simple comparison table/diagram
2. 7 Marks
   bExplain in detail about of Artificial Intelligence, Machine Learning and Deep Learning.
   How to answer

   • Define Artificial Intelligence as broad field
   • Define Machine Learning as subset of AI
   • Define Deep Learning as subset of ML using deep neural networks
   • Relationship diagram AI > ML > DL with examples
3. 8 Marks
   aWhat is Hierarchical Clustering? Explain DIANA clustering method.
   How to answer

   • Define hierarchical clustering and its two types agglomerative vs divisive
   • Explain DIANA as divisive top-down approach
   • Steps: start with one cluster, split by diameter, repeat
   • Advantages/limitations
4. 6 Marks
   bDiscuss different application areas where Clustering is used?
   How to answer

   • Define clustering purpose
   • List application domains: marketing, biology, image, document, anomaly detection etc.
   • Brief use in each domain
   • Summary of why clustering suits unlabeled data
5. 9 Marks
   aIs K-NN algorithm is supervised. Explain in detail.
   How to answer

   • State K-NN is supervised lazy instance-based learner
   • Explain training with labeled data and distance measures
   • Explain prediction by majority vote with example for K
   • Discuss choice of K, pros and cons
6. 5 Marks
   bWrite a note on Classification Accuracy.
   How to answer

   • Define classification accuracy
   • Give formula: (TP+TN)/total
   • Explain use with confusion matrix
   • Discuss limitations with imbalanced data
7. 7 Marks
   aHow is a Random Forest related to Decision Trees? Discuss.
   How to answer

   • Define decision tree as base learner
   • Define random forest as ensemble of trees via bagging
   • Explain random feature sampling and voting
   • Benefits: reduced overfitting, improved generalization
8. 7 Marks
   bWhat is Bagging and Boosting? Write few differences between them in detail.
   How to answer

   • Define bagging: bootstrap aggregating parallel
   • Define boosting: sequential error correction
   • Compare on training, bias-variance, examples Random Forest vs AdaBoost
   • Tabular differences
9. 7 Marks
   aWhat is dimensionality reduction? Explain the benefits of applying dimensionality reduction.
   How to answer

   • Define dimensionality reduction and curse of dimensionality
   • Explain need to reduce features
   • List benefits: less computation, avoids overfitting, visualization
   • Example PCA
10. 7 Marks
    bExplain Backward Elimination technique in detail.
    How to answer

    • Define backward elimination as wrapper feature selection
    • Steps: start with all features, remove least significant by p-value/performance
    • Stopping criterion
    • Pros/cons and example
11. 6 Marks
    aList out main approaches for dimensionality reduction.
    How to answer

    • Define two families: feature selection vs feature extraction
    • List selection: filter, wrapper, embedded
    • List extraction: PCA, LDA, t-SNE, autoencoders
    • One-line purpose of each
12. 8 Marks
    bDiscuss how Random Forest algorithm give output for Regression problems?
    How to answer

    • Recall random forest ensemble of trees
    • Explain each tree predicts continuous value
    • Explain aggregation by averaging for regression
    • Example and difference from majority voting in classification
13. 7 Marks
    aWhat is clustering? Discuss the importance of Expectation Maximization clustering method.
    How to answer

    • Define clustering as unsupervised grouping
    • Introduce EM for Gaussian mixtures
    • Explain E-step and M-step iteration
    • Importance: soft clustering, missing data, probabilistic basis
14. 7 Marks
    bDefine data science. List and explain different application areas of data science.
    How to answer

    • Define data science interdisciplinary field
    • List areas: healthcare, finance, e-commerce, social media etc.
    • Explain use in 3-4 areas
    • Conclusion on decision making
15. 14 Marks
    Write a short note on any two of the following:

    1. a)BIRCH Algorithm
    2. b)Confusion Matrix
    3. c)Data Augmentation
    4. d)Stacking
    How to answer

    • Select any two topics from four options
    • For each: definition, working steps, diagram/formula
    • Applications and advantages/limitations
    • Concluding comparison