MCQs

Disclaimer:

Educational Use Only:

The website links provided are to be used strictly for educational and knowledge-gaining purposes only.

No Misuse:

Do not misuse the links or the information contained therein for any commercial, illegal, or unethical activities.

Respect for Websites and Businesses:

Respect the intellectual property and business interests of the websites and their owners. Do not attempt to alter, copy, or redistribute content without proper authorization.

Acknowledge Value of Knowledge:

Recognize and value the knowledge and resources shared by these websites. Use the information responsibly and ethically.

No Endorsement:

The provision of these links does not imply endorsement of the linked sites, their content, or their products and services. Users are responsible for evaluating the suitability and accuracy of the information.

User Responsibility:

By accessing these links, you agree to use the information at your own risk and to comply with all applicable laws and website terms of use.

Overall MCQs for Reference

1. Which method is described as using a straight line to estimate a relationship between variables?
   a. Polynomial regression
   b. Linear regression
   c. Decision tree
   d. Neural network
   Answer: b. Linear regressionscribd​
2. Which algorithm provides a polynomial time 2-approximation for the Vertex Cover problem?
   a. Greedy Set Cover
   b. Linear Programming Rounding
   c. Brute force
   d. Dynamic Programming
   Answer: b. Linear Programming Roundingtestbook​
3. What is the primary advantage of approximation algorithms for NP-complete optimization problems?
   a. They always give the exact solution
   b. They never fail
   c. They produce near-optimal solutions in polynomial time
   d. They use brute-force techniques
   Answer: c. They produce near-optimal solutions in polynomial timegeeksforgeeks+1​
4. What technique is used to transform the state space in function approximation for reinforcement learning?
   a. One-hot encoding
   b. Basis function methods
   c. PCA
   d. Regularization
   Answer: b. Basis function methodsgeeksforgeeks​
5. Which theorem discusses the ability of neural networks to approximate any continuous function?
   a. Universal Approximation Theorem
   b. NP-Hardness Theorem
   c. Central Limit Theorem
   d. Taylor Theorem
   Answer: a. Universal Approximation Theoremwikipedia​
6. What is one advantage of locally weighted regression in machine learning?
   a. Ignores older data
   b. Pointwise approximation of complex target function
   c. Requires no training data
   d. Always overfits
   Answer: b. Pointwise approximation of complex target functiondiptisurvase.wordpress​
7. Neural networks are described as what type of function approximator in machine learning?
   a. Linear
   b. Non-linear
   c. Random
   d. Constant
   Answer: b. Non-linearscribd+1​
8. Stochastic approximation algorithms are primarily used for?
   a. Feature selection
   b. Finding zeroes or optimal points of functions under noise
   c. Data normalization
   d. Clustering
   Answer: b. Finding zeroes or optimal points of functions under noiseonlinecourses.nptel​
9. Which method uses kernels for non-parametric function approximation?
   a. Radial Basis Function networks
   b. Gaussian Processes
   c. Support Vector Machines
   d. Linear Regression
   Answer: b. Gaussian Processesgeeksforgeeks​
10. In polynomial regression, the target function is approximated using:
    a. Step functions
    b. High-degree polynomials
    c. Single lines
    d. Discrete trees
    Answer: b. High-degree polynomialsscribd​
11. Decision trees approximate functions by:
    a. Linear combinations
    b. Applying decision rules in a tree-like structure
    c. Matrix decomposition
    d. Feature selection
    Answer: b. Applying decision rules in a tree-like structurescribd​
12. What is the main goal of approximation in machine learning?
    a. To reduce accuracy
    b. To find simple models that predict unseen data
    c. To increase computational complexity
    d. To memorize the training set
    Answer: b. To find simple models that predict unseen dataarxiv+1​
13. What is the main measure of an approximation’s success in supervised learning?
    a. High complexity
    b. Low loss
    c. High dimensionality
    d. Large training data
    Answer: b. Low lossscribd​
14. What is a common example of a basis function in approximation?
    a. Sinusoidal waves
    b. Radial Basis Function
    c. Linear regression
    d. Quadratic model
    Answer: b. Radial Basis Functiongeeksforgeeks​
15. In the context of NP-complete problems, why are approximation algorithms important?
    a. Exact solutions are always available
    b. Polynomial time solutions cannot guarantee optimality
    c. They skip computations
    d. They memorize answers
    Answer: b. Polynomial time solutions cannot guarantee optimalitytestbook+1​
16. Which method can model complex, non-linear relationships for function approximation?
    a. Neural networks
    b. SVM with linear kernel
    c. Linear Regression
    d. PCA
    Answer: a. Neural networksarxiv+1​
17. What does the universal approximation theorem apply to?
    a. Decision trees
    b. Single-layer feedforward neural networks
    c. Clustering models
    d. Support vector machines
    Answer: b. Single-layer feedforward neural networkswikipedia​
18. Kernel methods like SVMs use what concept for function approximation?
    a. Feature expansion via kernel trick
    b. Feature reduction
    c. Clustering
    d. Naive Bayes theorem
    Answer: a. Feature expansion via kernel trickgeeksforgeeks​
19. Which loss function is commonly minimized during approximation in regression?
    a. Cross-entropy
    b. Mean squared error
    c. Hinge loss
    d. F1 score
    Answer: b. Mean squared errorscribd​
20. The curse of dimensionality is alleviated in high-dimensional PDE approximation by which method?
    a. Randomized neural networks
    b. Decision trees
    c. Brute force search
    d. Linear regression
    Answer: a. Randomized neural networksarxiv​
21. Approximating the value function in reinforcement learning can be done with:
    a. Linear functions
    b. Neural networks
    c. Basis functions
    d. All of the above
    Answer: d. All of the abovegeeksforgeeks​
22. What is the main computational advantage of function approximation in machine learning?
    a. It increases resource usage
    b. Enables efficient predictions with limited memory
    c. Reduces interpretability
    d. Increases complexity
    Answer: b. Enables efficient predictions with limited memoryarxiv+1​
23. What do kernel-based approximation methods offer in terms of function flexibility?
    a. Fixed form
    b. High flexibility and robustness
    c. Random predictions
    d. Computational simplicity
    Answer: b. High flexibility and robustnessgeeksforgeeks​
24. In neural network approximators, what are typically tuned during training?
    a. Sample weights
    b. Model biases
    c. Parameters like weights and biases
    d. Learning rate only
    Answer: c. Parameters like weights and biasesscribd​
25. What is a disadvantage of complex non-linear approximators like neural networks?
    a. Poor scalability to big data
    b. Need for substantial computational resources and careful tuning
    c. Lack of expressiveness
    d. Guaranteed overfitting
    Answer: b. Need for substantial computational resources and careful tuningarxiv+1​
26. Which concept describes expressing knowledge as conditional rules in models like decision trees?
    a. Linear regression
    b. Decision rules
    c. Gaussian processes
    d. Kernel expansion
    Answer: b. Decision rulesscribd​
27. Function approximation in reinforcement learning serves to:
    a. Generalize from training examples
    b. Memorize all state values
    c. Ignore unseen data
    d. Limit model complexity
    Answer: a. Generalize from training examplesincompleteideas+1​
28. The approximation error is defined as:
    a. The amount by which the approximating model differs from the target function
    b. Training time
    c. The optimal value
    d. Number of features
    Answer: a. The amount by which the approximating model differs from the target functionarxiv​
29. What does a “black box function” refer to in approximation models?
    a. Models with easily interpretable internal logic
    b. Highly complex models with inaccessible internal structure
    c. Only linear models
    d. Tree-based models
    Answer: b. Highly complex models with inaccessible internal structurescribd​
30. In function approximation, what structure is visually represented as graphs or diagrams?
    a. Decision trees
    b. Linear regression
    c. Neural networks
    d. k-NN classifiers
    Answer: a. Decision treesscribd​
31. What is often used to improve interpretability in tree-based approximators?
    a. Feature engineering
    b. Pruning
    c. Regularization
    d. Ensemble methods
    Answer: b. Pruningscribd​
32. Which method can be used for both regression and classification approximation tasks?
    a. Neural networks
    b. Linear regression only
    c. PCA
    d. Histogram equalization
    Answer: a. Neural networksarxiv+1​
33. What is the role of the loss function in model training for approximation?
    a. To increase training data
    b. To reduce approximation error by guiding parameter adjustment
    c. To increase error
    d. To select features
    Answer: b. To reduce approximation error by guiding parameter adjustmentscribd​
34. Which concept describes using subsets of data to estimate model accuracy in approximation?
    a. Cross-validation
    b. Kernel expansion
    c. Black-box testing
    d. Pruning
    Answer: a. Cross-validationscribd​
35. What is evaluated during bias-variance decomposition for approximation models?
    a. Model interpretability
    b. Trade-off between error due to bias and variance
    c. Feature importance
    d. Regularization strength
    Answer: b. Trade-off between error due to bias and variancearxiv​
36. Which method offers a non-parametric way to approximate functions using data similarity?
    a. Gaussian process regression
    b. Linear regression
    c. Polynomial regression
    d. Neural networks
    Answer: a. Gaussian process regressiongeeksforgeeks​
37. Approximation on manifolds is discussed as an emerging trend in:
    a. Deep learning architectures
    b. Support vector machines
    c. Ensemble methods
    d. Clustering
    Answer: a. Deep learning architecturesarxiv​
38. What property describes neural networks’ ability to fit a range of functions?
    a. Universal approximation property
    b. Decision tree flexibility
    c. Kernel method robustness
    d. Polynomial regression fit
    Answer: a. Universal approximation propertywikipedia+1​
39. What is a trade-off often encountered in approximation methods for large datasets?
    a. Speed vs. memory
    b. Complexity vs. accuracy
    c. Optimization vs. overfitting
    d. All of the above
    Answer: d. All of the abovearxiv+1​
40. The difference between training loss and actual test loss is largely due to:
    a. Overfitting
    b. Underfitting
    c. Kernel expansion
    d. Regularization
    Answer: a. Overfittingscribd​
41. What is the purpose of regularization in approximation models?
    a. Reduce model bias
    b. Control overfitting
    c. Prune decision trees
    d. Increase data size
    Answer: b. Control overfittingarxiv+1​
42. Non-linear approximation is especially important for:
    a. Simple datasets
    b. Complex datasets
    c. Small numbers
    d. High noise levels
    Answer: b. Complex datasetsgeeksforgeeks+1​
43. Model interpretability in approximation methods is improved by:
    a. Using simpler models
    b. Increasing layers in neural networks
    c. Randomizing outputs
    d. Ignoring feature importance
    Answer: a. Using simpler modelsscribd​
44. What is the main limitation of basis function methods as approximators?
    a. Scalability to high-dimensional spaces
    b. Lack of flexibility
    c. High memory usage
    d. Always overfit
    Answer: a. Scalability to high-dimensional spacesgeeksforgeeks​
45. Which term describes the set of model parameters in linear regression approximation?
    a. Decision rules
    b. Weights and biases
    c. Kernel maps
    d. Principal components
    Answer: b. Weights and biasesscribd​
46. Which application is cited for physics-informed approximation using neural surrogates?
    a. High-dimensional PDEs
    b. Image classification
    c. Text sentiment analysis
    d. Clustering
    Answer: a. High-dimensional PDEsarxiv​
47. Approximation algorithms for optimization guarantee:
    a. Exact solution
    b. Near-optimal solution within a proven bound
    c. Randomized solutions
    d. No error
    Answer: b. Near-optimal solution within a proven boundgeeksforgeeks+1​
48. Which basis function method divides the state space into tiles for easier function approximation?
    a. Polynomial basis
    b. Tile coding
    c. Neural network layers
    d. Principal components
    Answer: b. Tile codinggeeksforgeeks​
49. Why are randomized methods in neural networks effective for high-dimensional problems?
    a. Curse of dimensionality
    b. Dimension-independent convergence rates
    c. Slow computation
    d. Data reduction
    Answer: b. Dimension-independent convergence ratesarxiv​
50. Which approach in machine learning leverages approximation on manifolds for improved results?
    a. Shallow networks
    b. Deep learning networks
    c. SVM with linear kernel
    d. PCA-based models
    Answer: b. Deep learning networksarxiv​

1. https://www.scribd.com/document/882054141/Function-Approximation
2. https://testbook.com/question-answer/consider-the-following-statements-of-approximation–63e4ea683cdd1744cc1e3729
3. https://www.geeksforgeeks.org/dsa/approximation-algorithms/
4. https://www.geeksforgeeks.org/machine-learning/function-approximation-in-reinforcement-learning/
5. https://en.wikipedia.org/wiki/Universal_approximation_theorem
6. https://diptisurvase.wordpress.com/wp-content/uploads/2022/09/ml-mcqs.pdf
7. https://onlinecourses.nptel.ac.in/noc25_cs111/preview
8. https://arxiv.org/pdf/2506.02168.pdf
9. https://arxiv.org/html/2501.12145v1
10. http://incompleteideas.net/book/ebook/node85.html
11. https://testbook.com/objective-questions/mcq-on-errors-and-approximations–5eea6a1439140f30f369f15b
12. https://www.youtube.com/watch?v=hQf0bcwIYx8
13. https://www.sanfoundry.com/1000-machine-learning-questions-answers/
14. https://www.gowithquiz.com/unit-1-mcq-machine-learning
15. https://www.scribd.com/document/903942674/Ml-Mcq-All-5-Machine-Learning-Mcqs
16. https://www.sciencedirect.com/science/article/pii/S0278262615300038
17. https://npfsa2017.uni-jena.de/l_notes/vybiral.pdf
18. https://codecrucks.com/mcq/machine-learning-mcqs-set-05/
19. https://nadavdym.github.io/lecture_notes/DeepNotes.pdf
20. https://www.coursehero.com/file/183492221/MCQs-approximation-algorithmspdf/