Question Bank
GATE ML Practice Questions
150 carefully crafted, exam-pattern questions across all 10 Machine Learning syllabus units with complete worked mathematical solutions and distractor explanations.
→ Launch Interactive Practice App
1. About the Question Bank
The GATE ML question bank provides focused, high-yield drill problems aligned with the official GATE Data Science & Artificial Intelligence (DA) examination syllabus. Unlike generic online quizzes that only test surface terminology, our questions emphasize derivation, parameter sensitivity, matrix computations, and loss function mechanics.
2. Unit-Wise Distribution
The 150 questions are distributed across the complete syllabus to guarantee thorough coverage:
| Guide Unit | Topic Domain | Key Tested Mechanics |
|---|---|---|
| Unit 1 | Foundations & Maximum Likelihood | Empirical Risk Minimization, Maximum Likelihood Estimation (MLE), Maximum A Posteriori (MAP). |
| Unit 2 | Linear & Polynomial Regression | Ordinary Least Squares (OLS) closed form, Moore-Penrose pseudoinverse, Gradient Descent convergence. |
| Unit 3 | Ridge & Lasso Regularization | L2 weight shrinkage, L1 sparsity geometry, subgradient optimization, bias-variance trade-off. |
| Unit 4 | Logistic Regression | Sigmoid properties, cross-entropy loss convex optimization, decision boundary geometry. |
| Unit 5 | k-Nearest Neighbors (k-NN) | Distance metrics (Euclidean, Manhattan, Minkowski), curse of dimensionality, boundary smoothness vs k. |
| Unit 6 | Naive Bayes Classifier | Class-conditional independence, Laplace/Lidstone smoothing, zero-frequency mitigation. |
| Unit 7 | Linear Discriminant Analysis | Fisher criterion, between-class vs within-class scatter matrices, projection directions. |
| Unit 8 | Support Vector Machines | Hard vs soft margin, slack variables, dual Lagrangian formulation, kernel trick (RBF, Polynomial). |
| Unit 9 | Decision Trees & Ensembles | Information Gain (Shannon entropy), Gini impurity, CART variance reduction, pruning rules. |
| Unit 10 | Clustering & Dimensionality Reduction | k-Means centroid updates, inertia convergence, PCA covariance eigen-decomposition, explained variance. |
3. Question Formats & Real Exam Marking Rules
Every drill session enforces the exact scoring rules used by the Indian Institutes of Technology (IITs):
- Multiple Choice (MCQ): Four choices with a single correct option. 1-mark questions carry a penalty of
-⅓(-0.33) marks for wrong answers; 2-mark questions carry a penalty of-⅔(-0.67) marks. Unattempted questions incur 0 marks. - Multiple Select (MSQ): Four choices where one, two, three, or all four options may be correct. Zero negative marking, but no partial credit. You must identify all correct statements and avoid any distractor to receive full marks.
- Numerical Answer Type (NAT): Enter exact decimal answers via on-screen numerical keypad. All solutions specify tolerance windows (e.g.,
[1.41, 1.43]). There is no negative marking on NAT questions.
4. Keyboard Shortcuts for Power Revision
Our practice interface includes hotkeys designed for keyboard-driven review:
A,B,C,Dor1,2,3,4: Select optionsEnter: Submit and verify responseN/P: Next / Previous questionM: Mark / unmark for reviewF: Toggle difficulty flag
5. Next Steps
After completing your practice drills, review your performance in the Timed Mock Exam or study theoretical derivations in the Machine Learning Study Guide.