Scribd
Upload
4 views4 pages

Deep Learning

The document contains a series of questions and answers related to deep learning concepts, including characteristics of neural networks, activation functions, and specific models like CNNs and RNNs. Key topics covered include the purpose of dropout, convolution operations, the vanishing gradient problem, and transfer learning. Additionally, it discusses optimizers that adapt learning rates during training.

Uploaded by

anurandey5
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
4 views4 pages

Deep Learning

The document contains a series of questions and answers related to deep learning concepts, including characteristics of neural networks, activation functions, and specific models like CNNs and RNNs. Key topics covered include the purpose of dropout, convolution operations, the vanishing gradient problem, and transfer learning. Additionally, it discusses optimizers that adapt learning rates during training.

Uploaded by

anurandey5
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

1. What is a key characteristic of deep learning?

A. Uses decision trees for prediction​


B. Involves manual feature extraction​
C. Uses multi-layered neural networks​
D. Requires less data than traditional machine learning​
Answer: C. Uses multi-layered neural networks

2. What is the purpose of an activation function in a neural network?

A. To normalize the output​


B. To make the model linear​
C. To introduce non-linearity​
D. To initialize weights​
Answer: C. To introduce non-linearity

3. Which of the following is a commonly used activation function?

A. Dropout​
B. ReLU​
C. Softmax Regression​
D. Stochastic Gradient Descent​
Answer: B. ReLU

4. What does ReLU stand for?

A. Rectified Linear Unit​


B. Regularized Linear Unit​
C. Recurrent Linear Update​
D. Restricted Learning Unit​
Answer: A. Rectified Linear Unit

5. Which of the following deep learning models is best suited for sequential
data?
A. Convolutional Neural Networks (CNN)​
B. Recurrent Neural Networks (RNN)​
C. Random Forest​
D. Logistic Regression​
Answer: B. Recurrent Neural Networks (RNN)

6. What is the main purpose of dropout in deep learning?

A. To increase training speed​


B. To reduce overfitting​
C. To enhance activation functions​
D. To increase data size​
Answer: B. To reduce overfitting

7. Which layer is used in CNNs to reduce spatial dimensions?

A. Fully Connected Layer​


B. Dropout Layer​
C. Pooling Layer​
D. ReLU Layer​
Answer: C. Pooling Layer

8. What does a convolution operation do in CNNs?

A. Flattens the input​


B. Applies filters to extract features​
C. Normalizes the output​
D. Connects all neurons​
Answer: B. Applies filters to extract features

9. In deep learning, backpropagation is used to:

A. Initialize weights​
B. Propagate input forward​
C. Update weights based on error​
D. Create new layers​
Answer: C. Update weights based on error
10. What is the vanishing gradient problem?

A. When gradients become too large​


B. When gradients become too small during backpropagation​
C. When training stops unexpectedly​
D. When a model overfits​
Answer: B. When gradients become too small during backpropagation

11. Which type of neural network is best for image recognition tasks?

A. RNN​
B. LSTM​
C. CNN​
D. GAN​
Answer: C. CNN

12. What is the function of the Softmax layer in a neural network?

A. To add regularization​
B. To output probabilities for multi-class classification​
C. To reduce dimensionality​
D. To increase learning rate​
Answer: B. To output probabilities for multi-class classification

13. What is transfer learning?

A. Training a new model from scratch​


B. Using weights from a pretrained model on a new problem​
C. Discarding old training data​
D. Manually tuning hyperparameters​
Answer: B. Using weights from a pretrained model on a new problem

14. What is a Generative Adversarial Network (GAN)?


A. A network used for classification​
B. A pair of networks competing against each other to generate data​
C. A type of RNN​
D. A network that performs dimensionality reduction​
Answer: B. A pair of networks competing against each other to generate data

15. Which optimizer adapts the learning rate for each parameter during
training?

A. SGD​
B. Adam​
C. RMSprop​
D. Both B and C​
Answer: D. Both B and C

You might also like