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. What is deep learning?
   a. A type of machine learning algorithm
   b. A branch of artificial intelligence
   c. A technique for training neural networks
   d. All of the above
   Answer: d. All of the aboveaimcqs​
2. Which of the following is not a popular deep learning framework?
   a. TensorFlow
   b. PyTorch
   c. Keras
   d. Scikit-learn
   Answer: d. Scikit-learnaimcqs​
3. What is the purpose of an activation function in a neural network?
   a. Introduces non-linearity to the network
   b. Determines neuron output
   c. Helps in backpropagation
   d. All of the above
   Answer: d. All of the aboveaimcqs​
4. What is the role of gradient descent in deep learning?
   a. Minimize the loss function
   b. Find the optimal weights
   c. Update network parameters
   d. All of the above
   Answer: d. All of the aboveaimcqs​
5. What is the vanishing gradient problem?
   a. Gradients become extremely small
   b. Gradients become extremely large
   c. Gradients remain unchanged
   d. None of the above
   Answer: a. Gradients become extremely smallscribd+1​
6. What is backpropagation?
   a. Forward pass method
   b. Weight adjustment method based on errors
   c. Prevents overfitting
   d. Feature selection method
   Answer: b. Weight adjustment method based on errorsscribd​
7. Which activation function is commonly used in hidden layers?
   a. Softmax
   b. ReLU
   c. Sigmoid
   d. Tanh
   Answer: b. ReLUscribd​
8. Which activation function is commonly used for binary classification output layers?
   a. Sigmoid
   b. ReLU
   c. Tanh
   d. Softmax
   Answer: a. Sigmoidscribd​
9. What is the main use of convolutional layers in CNNs?
   a. Maintains spatial hierarchies
   b. Performs mathematical multiplication
   c. Performs classification
   d. None of the above
   Answer: a. Maintains spatial hierarchiesletsfindcourse+1​
10. What is the major purpose of pooling layers in CNN?
    a. Reduces overfitting
    b. Downsamples feature maps
    c. Increases computation
    d. Upsamples input
    Answer: b. Downsamples feature mapsscribd​
11. Which neural network is primarily used for image data?
    a. RNN
    b. CNN
    c. DBN
    d. Autoencoder
    Answer: b. CNNscribd​
12. Which deep learning technique is used for sequence data like language or time-series?
    a. CNN
    b. DBN
    c. RNN
    d. GAN
    Answer: c. RNNscribd​
13. Which deep learning technique generates new, realistic data samples?
    a. Transfer Learning
    b. Reinforcement Learning
    c. GAN
    d. LSTM
    Answer: c. GANscribd​
14. What is transfer learning used for?
    a. Training on small data
    b. Reusing a pre-trained model
    c. Data augmentation
    d. Optimizing loss
    Answer: b. Reusing a pre-trained modelscribd​
15. Which layer connects every neuron in one layer to every neuron in the next?
    a. Pooling layer
    b. Convolutional layer
    c. Fully connected layer
    d. Batch normalization layer
    Answer: c. Fully connected layerscribd​
16. What does dropout do in a neural network?
    a. Increases input size
    b. Randomly ignores neurons during training
    c. Changes learning rate
    d. Increases overfitting
    Answer: b. Randomly ignores neurons during trainingscribd​
17. Which metric is commonly used to evaluate classification models in deep learning?
    a. R-squared
    b. Mean squared error
    c. Accuracy
    d. Correlation
    Answer: c. Accuracyscribd​
18. What is the main idea of reinforcement learning?
    a. Learning from labeled data
    b. Learning from delayed rewards
    c. Learning from clustering
    d. None of the above
    Answer: b. Learning from delayed rewardsscribd+1​
19. Which architecture is commonly used for detecting objects in images?
    a. CNN
    b. RNN
    c. DBN
    d. SVM
    Answer: a. CNNscribd​
20. Which function is used in output layers for multi-class classification?
    a. Sigmoid
    b. ReLU
    c. Tanh
    d. Softmax
    Answer: d. Softmaxscribd​
21. Which technique is applied to prevent overfitting?
    a. Increasing epochs
    b. Decreasing features
    c. Regularization
    d. Adding layers
    Answer: c. Regularizationscribd​
22. LSTM is a specialized form of which neural network?
    a. CNN
    b. DBN
    c. RNN
    d. GAN
    Answer: c. RNNscribd​
23. What does batch normalization help to solve?
    a. Underfitting
    b. Exploding/vanishing gradients
    c. Overfitting
    d. High variance
    Answer: b. Exploding/vanishing gradientsscribd​
24. Which algorithm adapts learning rates based on gradients?
    a. Adagrad
    b. Adam
    c. RMSprop
    d. All of the above
    Answer: d. All of the abovescribd​
25. Which optimization technique adds “momentum” to help with faster convergence?
    a. Adam
    b. Momentum-based SGD
    c. Nesterov Accelerated Gradient
    d. All of the above
    Answer: d. All of the abovescribd​
26. Q-learning belongs to which category?
    a. Classification
    b. Unsupervised learning
    c. Value-based reinforcement learning
    d. Clustering
    Answer: c. Value-based reinforcement learningscribd​
27. What is model distillation?
    a. Using smaller models with larger ones as guides
    b. Data augmentation
    c. Ensemble learning
    d. Unsupervised pre-training
    Answer: a. Using smaller models with larger ones as guidesscribd​
28. Which method provides local, explainable predictions in XAI?
    a. Deep Dream
    b. LIME
    c. One-hot encoding
    d. Data normalization
    Answer: b. LIMEscribd​
29. What does federated learning do?
    a. Transfers weights between networks
    b. Trains models on decentralized data with privacy
    c. Distills models
    d. Augments data
    Answer: b. Trains models on decentralized data with privacyscribd​
30. Which technique helps a model generalize to new data?
    a. Underfitting
    b. Data augmentation
    c. Reducing epochs
    d. Increasing loss
    Answer: b. Data augmentationscribd​
31. What is a typical cause of exploding gradient?
    a. Small learning rates
    b. Very deep networks
    c. Too many output neurons
    d. Large batch sizes
    Answer: b. Very deep networksscribd+1​
32. Which is NOT a common activation function in deep learning?
    a. ReLU
    b. Tanh
    c. Linear
    d. Step
    Answer: d. Stepscribd​
33. Autoencoders are primarily used for:
    a. Time series prediction
    b. Image classification
    c. Unsupervised feature learning
    d. Text embedding
    Answer: c. Unsupervised feature learningscribd​
34. Shallow neural networks have:
    a. One or two hidden layers
    b. Many hidden layers
    c. More neurons than deep networks
    d. None of the above
    Answer: a. One or two hidden layersgeeksforgeeks​
35. Which technique penalizes large weights in the loss function?
    a. Dropout
    b. L1 Regularization
    c. Batch Normalization
    d. L2 Regularization
    Answer: d. L2 Regularizationscribd​
36. GANs consist of:
    a. Generator and Discriminator
    b. Encoder and Decoder
    c. Two classifiers
    d. No trainable weights
    Answer: a. Generator and Discriminatorcs230.stanford​
37. What is the main goal of the Generator in a GAN?
    a. Distinguish fake vs. real data
    b. Generate data samples similar to real data
    c. Supervise the Discriminator
    d. Minimize accuracy
    Answer: b. Generate data samples similar to real datacs230.stanford​
38. What technique randomly selects neurons to ignore during training?
    a. L2 Regularization
    b. Dropout
    c. Pooling
    d. Adam
    Answer: b. Dropoutscribd​
39. Which layer helps networks learn translation invariance in images?
    a. Fully connected
    b. Pooling
    c. Batch normalization
    d. Skip
    Answer: b. Poolingscribd​
40. Which optimizer is known for fast convergence in deep nets?
    a. Adam
    b. Vanilla SGD
    c. Adagrad
    d. Newton-Raphson
    Answer: a. Adamscribd​
41. What is the ‘filter’ in a convolutional layer?
    a. A mathematical constant
    b. A window of weights sliding across input
    c. An output map
    d. An input neuron
    Answer: b. A window of weights sliding across inputscribd​
42. In CNN, 1×1 convolutions are used to:
    a. Increase spatial dimensions
    b. Act as feature pooling
    c. Reduce depth
    d. Upsample images
    Answer: c. Reduce depthletsfindcourse​
43. Which deep learning method uses attention mechanisms for sequence-to-sequence tasks?
    a. RNN
    b. Transformer
    c. CNN
    d. Autoencoder
    Answer: b. Transformerscribd+1​
44. Skip connections are used in which architecture?
    a. LeNet
    b. ResNet
    c. VGGNet
    d. AlexNet
    Answer: b. ResNetgtbit​
45. What is the key difference between LSTM and GRU?
    a. GRU has less gating mechanisms than LSTM
    b. LSTM stores no state
    c. LSTM is faster to train
    d. GRU cannot handle long-term dependencies
    Answer: a. GRU has less gating mechanismsscribd​
46. Which metric is best for imbalanced classification tasks?
    a. Mean Absolute Error
    b. Accuracy
    c. Precision-Recall/AUC
    d. R-squared
    Answer: c. Precision-Recall/AUCscribd​
47. Deep Belief Networks (DBN) are stacks of:
    a. Perceptrons
    b. Restricted Boltzmann Machines
    c. Autoencoders
    d. SVMs
    Answer: b. Restricted Boltzmann Machinesscribd​
48. What is required for supervised deep learning?
    a. Unlabeled data
    b. Labeled data
    c. Reinforcement signal
    d. No input data
    Answer: b. Labeled datascribd​
49. Which deep learning technique is useful for dimensionality reduction?
    a. GAN
    b. PCA
    c. Autoencoder
    d. Random Forest
    Answer: c. Autoencoderscribd​
50. What is the main limitation of deep learning?
    a. Cannot learn from raw data
    b. Requires a lot of labeled data and computation
    c. No need for hyperparameter tuning
    d. Works only on tabular data
    Answer: b. Requires a lot of labeled data and computationgeeksforgeeks+1​

1. https://aimcqs.com/fundamentals
2. https://www.scribd.com/document/851342076/Deep-Learning-117-MCQ
3. https://letsfindcourse.com/data-science/deep-learning-mcq-questions-and-answers
4. https://www.scribd.com/document/865132941/Exam-Practice-Questions
5. https://www.geeksforgeeks.org/deep-learning/deep-learning-interview-questions/
6. https://cs230.stanford.edu/files/cs230exam_win19_soln.pdf
7. https://www.gtbit.ac.in/assets/img/pdf/ADL%20Question%20bank.pdf
8. https://www.scribd.com/document/488694104/Deep-Learning-MCQ
9. https://www.gyansetu.in/blog/60-deep-learning-objective-type-quiz-mcqs-with-answers-for-interview-preparation/
10. https://www.sanfoundry.com/1000-machine-learning-questions-answers/
11. https://www.acmecollinsschool.com/deep-learning-mcqs-questions.html
12. https://www.scribd.com/document/819504690/Deep-Learning-Question-Papers
13. https://www.interviewbit.com/machine-learning-mcq/
14. https://www.sanfoundry.com/1000-neural-networks-questions-answers/
15. https://www.exampaperspractice.co.uk/wp-content/uploads/18.1-Artificial-Intelligence.pdf.pdf
16. https://www.youtube.com/watch?v=FugpqOEPuBg
17. https://www.linkedin.com/pulse/50-comprehensive-questions-answers-advanced-machine-yg67c
18. https://www.aionlinecourse.com/ai-quiz-questions/machine-learning/deep-learning-ml
19. https://testbook.com/objective-questions/mcq-on-artificial-intelligence–5eea6a0d39140f30f369e30e
20. https://www.youtube.com/watch?v=Mr1DKghufmk