wayground logo

Free Printable Worksheets

NEW

Font size

S
M
L
XL
Worksheets

Page 1

Total questions: 30

Worksheet time: 15mins

Name
Class
Date
1.

Which of the following best describes Deep Learning?

a)

Rule-based programming

b)

A subset of Machine Learning using shallow models

c)

A subset of Machine Learning using multi-layer neural networks

d)

A database technique

2.

Which biological structure inspired artificial neural networks?

a)

Heart

b)

Brain neurons

c)

Lungs

d)

DNA

3.

In a neural network, what does a neuron compute?

a)

Maximum value

b)

Weighted sum + bias

c)

Only multiplication

d)

Logical AND

4.

Which activation function outputs values between 0 and 1?

a)

ReLU

b)

Tanh

c)

Sigmoid

d)

Linear

5.

Which activation function helps reduce the vanishing gradient problem?

a)

Sigmoid

b)

Tanh

c)

ReLU

d)

Softmax

6.

What is the main objective of gradient descent?

a)

Increase accuracy

b)

Minimize loss function

c)

Increase learning rate

d)

Maximize weights

7.

Which loss function is commonly used for regression problems?

a)

Cross-entropy

b)

Mean Squared Error

c)

Hinge loss

d)

Log loss

8.

What does pooling mainly do in CNNs?

a)

Increase parameters

b)

Add non-linearity

c)

Reduce spatial dimensions

d)

Increase feature maps

9.

CNNs are mainly designed to work with which type of data?

a)

Tabular

b)

Sequential

c)

Spatial

d)

Text-only

10.

Which optimizer combines momentum and adaptive learning rates?

a)

SGD

b)

RMSProp

c)

Adam

d)

Adagrad

11.

What problem does backpropagation solve?

a)

Feature extraction

b)

Weight initialization

c)

Gradient computation

d)

Data normalization

12.

Which technique helps prevent overfitting by randomly disabling neurons?

a)

Regularization

b)

Dropout

c)

Pooling

d)

Padding

13.

What is the role of padding in CNNs?

a)

Increase filter size

b)

Preserve spatial dimensions

c)

Increase stride

d)

Reduce channels

14.

Which CNN architecture first introduced deep CNNs with ReLU and GPU training?

a)

LeNet

b)

VGG

c)

AlexNet

d)

ResNet

15.

In CNNs, what does the number of filters represent?

a)

Number of neurons

b)

Number of feature maps

c)

Image size

d)

Number of classes

16.

Which architecture introduced skip connections?

a)

VGG

b)

ResNet

c)

AlexNet

d)

Inception

17.

Why is SAME padding important in ResNet?

a)

To reduce computation

b)

To increase filter size

c)

To match dimensions for addition

d)

To reduce parameters

18.

What is the purpose of the softmax function?

a)

Reduce loss

b)

Normalize gradients

c)

Convert outputs to probabilities

d)

Prevent overfitting

19.

Which optimizer uses a moving average of squared gradients?

a)

SGD

b)

Momentum

c)

RMSProp

d)

Adam

20.

Transfer learning mainly helps when:

a)

Dataset is extremely large

b)

Dataset is small or medium

c)

Model is shallow

d)

No GPU is available

21.

Why are small 3×3 filters preferred in VGG networks?

a)

Reduce stride

b)

Reduce depth

22.

What happens if learning rate is too high?

a)

Slow convergence

b)

No learning

c)

Overshooting the minimum

d)

Better accuracy

23.

In L2 regularization, which term is added to the loss function?

a)

Sum of weights

b)

Sum of squared weights

c)

Absolute weights

d)

Bias only

24.

In a CNN, why is max pooling commonly used after convolution layers?

a)

To increase the number of parameters

b)

To reduce spatial size and computation

c)

To add more filters

d)

To normalize pixel values

25.

What is the main advantage of residual learning?

a)

Reduces dataset size

b)

Eliminates pooling

c)

Improves gradient flow

d)

Removes fully connected layers

26.

Which of the following is NOT a solution to overfitting?

a)

Dropout

b)

Data augmentation

c)

Increasing model complexity

d)

Regularization

27.

Why does CNN use shared weights?

a)

Increase memory usage

b)

Reduce number of parameters

c)

Increase image size

d)

Increase overfitting

28.

What is the role of momentum in gradient descent?

a)

Increase batch size

b)

Reduce loss function

c)

Speed up convergence

d)

Increase overfitting

29.

In ResNet, what happens if the residual function learns zero mapping?

a)

The network stops training

b)

The output becomes zero

c)

The block behaves like an identity mapping

d)

The gradients vanish

30.

Which component of CNN performs final classification?

a)

Convolution layer

b)

Pooling layer

c)

Fully connected layer

d)

Padding layer