AL-405 Machine Learning - Jun 2026 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
   aExplain Machine Learning. Discuss its scope, limitations and real-world applications with suitable examples.
   How to answer

   • Definition of Machine Learning (Arthur Samuel / Tom Mitchell).
   • Scope: supervised, unsupervised, reinforcement, deep learning.
   • Limitations: data dependency, generalization, bias, interpretability, computation.
   • Real-world applications with examples e.g. spam detection, recommendation, healthcare.
2. 7 Marks
   bExplain Hypothesis Space and Inductive Bias in Machine Learning. Why are they important in model learning?
   How to answer

   • Definition of hypothesis space (set of all hypotheses).
   • Definition of inductive bias (assumptions for generalization).
   • Types of inductive bias: restriction, preference.
   • Importance in model learning: generalization, overfitting control, search efficiency.
3. 7 Marks
   aExplain Principal Component Analysis (PCA) with mathematical steps and advantages. Compare PCA with Partial Least Squares.
   How to answer

   • PCA concept: dimensionality reduction via variance maximization.
   • Mathematical steps: standardization, covariance, eigen decomposition, projection.
   • Advantages: noise reduction, visualization, compression.
   • Comparison with Partial Least Squares on supervision, objective, use-case.
4. 7 Marks
   bExplain the architecture of Multilayer Perceptron (MLP). Describe the Backpropagation Algorithm with a neat diagram.
   How to answer

   • MLP architecture: input, hidden, output layers, weights, activations.
   • Forward pass and loss computation.
   • Backpropagation steps: chain rule, gradient descent, weight update.
   • Neat labelled diagram of MLP.
5. 7 Marks
   aWhat are Vanishing Gradient and Exploding Gradient problems in detail?
   How to answer

   • Definition of vanishing gradient: gradients shrink in deep layers.
   • Definition of exploding gradient: gradients grow exponentially.
   • Causes: deep networks, sigmoid/tanh, weight initialization.
   • Effects and solutions: ReLU, batch norm, clipping, LSTM, initialization.
6. 7 Marks
   bDescribe the working of ID3 or CART algorithm with an example.
   How to answer

   • ID3/CART purpose: decision tree learning for classification/regression.
   • Splitting criteria: entropy/information gain or Gini index.
   • Algorithm steps with example dataset calculation.
   • Tree construction, pruning and example prediction.
7. 7 Marks
   aExplain Naive Bayes Classifier. State its assumptions, advantages, disadvantages and applications.
   How to answer

   • Bayes theorem and Naive independence assumption.
   • Classifier working: posterior calculation and decision rule.
   • Assumptions, advantages e.g. small data, fast.
   • Disadvantages, applications e.g. spam, sentiment analysis.
8. 7 Marks
   bExplain Random Forest Algorithm for both Classification and Regression.
   How to answer

   • Random Forest concept: bagging ensemble of decision trees.
   • Classification working: majority voting.
   • Regression working: averaging predictions.
   • Out-of-bag, feature randomness and advantages.
9. 7 Marks
   aWhat is Logistic Regression and Linear Regression with suitable examples.
   How to answer

   • Linear Regression definition, equation and example.
   • Logistic Regression definition, sigmoid and example.
   • Key differences: output type, function, loss.
   • Suitable examples for each.
10. 7 Marks
    bExplain Unsupervised Learning. Describe K-Means Clustering Algorithm with its advantages.
    How to answer

    • Unsupervised Learning definition and types.
    • K-Means steps: initialize, assign, update, converge.
    • Distance measure and choice of K.
    • Advantages: simple, scalable, applications.
11. 7 Marks
    aDifferentiate between Agglomerative and Divisive Clustering with examples.
    How to answer

    • Agglomerative: bottom-up merging definition and steps.
    • Divisive: top-down splitting definition and steps.
    • Differences on direction, complexity, accuracy.
    • Examples e.g. AGNES vs DIANA.
12. 7 Marks
    bExplain Gaussian Mixture Model (GMM). Discuss the Expectation-Maximization (EM) Algorithm with flowchart.
    How to answer

    • GMM concept: mixture of Gaussians for soft clustering.
    • Parameters: weights, means, covariances.
    • EM algorithm: E-step and M-step iteration.
    • Flowchart of EM for GMM.
13. 7 Marks
    aExplain the Design and Analysis of Machine Learning Experiments in detail.
    How to answer

    • Need for designed ML experiments.
    • Factors, response, training-test design.
    • Hypothesis, replication, randomization.
    • Analysis: performance estimation and comparison.
14. 7 Marks
    bExplain various Cross-Validation and Resampling Methods used in Machine Learning experiments.
    How to answer

    • Purpose of cross-validation and resampling.
    • Methods: holdout, k-fold, LOOCV, stratified, bootstrap.
    • Steps and pros-cons of each.
    • Use in model selection and error estimation.
15. 7 Marks
    aExplain Hypothesis Testing in Machine Learning in detail.
    How to answer

    • Hypothesis testing concept in ML model comparison.
    • Null/alternate hypothesis, p-value, significance.
    • Tests: t-test, paired test, McNemar, ANOVA.
    • Example of comparing two classifiers.
16. 7 Marks
    bDiscuss Accuracy, Precision, Recall and F1-Score.
    How to answer

    • Definitions from confusion matrix.
    • Formulas for Accuracy, Precision, Recall, F1-Score.
    • When each metric is useful, imbalance case.
    • Example calculation.