wayground logo

Free Printable Worksheets

NEW

Font size

S
M
L
XL
Worksheets

High School Level Neural Networks Quiz

Total questions: 50

Worksheet time: 25mins

Name
Class
Date
1.

In fully-connected layers, how does input pass through activation units?

a)

As individual pixels

b)

As a whole entity

c)

Through a floating window

d)

By specific pixel clusters

2.

Why are fully-connected layers often not the first choice for computational efficiency?

a)

They are too simple

b)

They require less memory

c)

They are computationally intensive

d)

They are less prone to overfitting

3.

What is a disadvantage of fully-connected layers regarding image shape?

a)

They are independent of image shape

b)

They improve with varying image shapes

c)

They become dependent on the shape of train images

d)

They maintain a fixed image shape

4.

How do fully-connected layers compare to single convolution operations in terms of parameters?

a)

Fully-connected layers have fewer parameters

b)

Single convolution operations increase parameters

c)

Fully-connected layers have a larger number of parameters

d)

Both have an equal number of parameters

5.

What can a Convolutional Neural Network (ConvNet) assign to various objects in an image?

a)

Random values

b)

Learnable weights and biases

c)

Fixed pixel colors

d)

Output predictions directly

6.

How does a ConvNet capture spatial and temporal dependencies in an image?

a)

By ignoring filters

b)

Through random sampling

c)

By applying relevant filters

d)

Using only fully-connected layers

7.

Which of the following is a component of a ConvNet architecture?

a)

Output layer only

b)

Input, Convolutional Layers, Pooling Layers, Fully-Connected Layers

c)

Activation units only

d)

Image processing software

8.

What happens when images reach higher dimensions, such as 8K images, in terms of computation?

a)

It becomes less intensive

b)

It remains unchanged

c)

It becomes computationally-intensive

d)

It speeds up processing

9.

What can ConvNet do to image dimensions without losing critical features?

a)

Increase them

b)

Maintain them

c)

Reduce them

d)

Distort them

10.

What are common filter (kernel) sizes for convolutional layers?

a)

1x1 or 2x2

b)

3x3 or 5x5

c)

7x7 or 9x9

d)

10x10 or larger

11.

What is the main objective of the convolution operation?

a)

To add noise to the image

b)

To reduce image brightness

c)

To extract high-level features like edges

d)

To flatten the image entirely

12.

How does the number of parameters in a convolutional layer compare to fully-connected layers?

a)

It is much more

b)

It is about the same

c)

It is much less

d)

It depends solely on the image size

13.

What issue can arise when applying many successive convolutional layers without padding?

a)

Image clarity increases

b)

Information in corner pixels can be lost

c)

Computational time decreases

d)

Feature extraction becomes less effective

14.

What is the solution to preserve the size of the original image during convolution and allow for more accurate analysis?

a)

Increasing filter size

b)

Zero padding

c)

Reducing image dimensions

d)

Using larger strides

15.

What does 'stride' control in a convolutional layer?

a)

The filter's color

b)

How the filter convolves through the input layer

c)

The image's resolution

d)

The number of output channels

16.

What is the default value for stride?

a)

0

b)

1

c)

2

d)

3

17.

How can a larger stride be used?

a)

To decrease computational efficiency

b)

To increase feature size

c)

To increase computational efficiency or down sample the feature size

d)

To add more information to the image

18.

In convolution operations, what must the channels of each kernel/filter equal?

a)

The number of output layers

b)

The channels of the input layer

c)

A fixed number, regardless of input

d)

The batch size

19.

Why might the number of filters be increased higher up in a neural network architecture?

a)

To make the network simpler

b)

To reduce computational load

c)

To extract more abstractions from the images

d)

To decrease accuracy

20.

What is the primary purpose of pooling layers?

a)

To increase the dimensions of feature maps

b)

To add more parameters to learn

c)

To reduce the dimensions of feature maps

d)

To make the model less robust

21.

What do pooling layers summarize?

a)

The raw input pixels

b)

The features present in a region of the feature map

c)

The entire dataset

d)

The final output probabilities

22.

How does pooling make the model more robust to variations in feature position?

a)

By using precisely positioned features

b)

By summarizing features instead of using precisely positioned features

c)

By adding more precise details

d)

By ignoring feature positions

23.

Which pooling method selects the maximum pixel value of the batch?

a)

Min pooling

b)

Average pooling

c)

Max pooling

d)

Sum pooling

24.

Which pooling method selects the minimum pixel value of the batch?

a)

Max pooling

b)

Average pooling

c)

Min pooling

d)

Median pooling

25.

Which pooling method selects the average value of all the pixels in the batch?

a)

Max pooling

b)

Average pooling

c)

Min pooling

d)

Weighted pooling

26.

In a modern CNN architecture, where do fully-connected layers typically form?

a)

The first few layers

b)

The middle layers

c)

The last few layers

d)

Only the input layer

27.

What happens to the output from the last convolutional/pooling layer before being fed into the fully-connected layer?

a)

It is inverted

b)

It is reshaped

c)

It is flattened

d)

It is duplicated

28.

What activation functions are typically used by the final layer of a CNN to calculate probabilities of different classes?

a)

ReLU or Tanh

b)

Sigmoid / Softmax

c)

Linear or Step

d)

Identity or Leaky ReLU

29.

What is the relationship between deep learning model performance and the amount of training data?

a)

Inversely proportional

b)

Not related

c)

Proportional

d)

Random

30.

What is 'Data Augmentation' used for?

a)

To decrease the amount of training data

b)

To invent more data

c)

To remove outliers from data

d)

To make data less diverse

31.

Which of the following is an example of Data Augmentation?

a)

Deleting data

b)

Randomizing labels

c)

Rotation, cropping, shifting

d)

Reducing resolution

32.

What is semi-supervised learning used for in improving performance?

a)

To eliminate labeled data

b)

To supplement training data with non-labelled data

c)

To increase the number of features

d)

To complicate the model architecture

33.

What is a Regression Algorithm a subset of?

a)

Unsupervised learning

b)

Reinforcement learning

c)

Supervised learning

d)

Deep learning

34.

What kind of value does the output variable in a regression algorithm predict?

a)

Categorical value

b)

Binary value

c)

Continuous/real value

d)

Discrete value

35.

What is one real-world scenario where regression is used?

a)

Image classification

b)

Voice recognition

c)

Weather condition prediction

d)

Spam detection

36.

What can regression help determine about factors?

a)

Only the least important factors

b)

Only the most important factors

c)

The most important and the least important factors

d)

No relationship between factors

37.

Which of the following is a type of Regression algorithm?

a)

Clustering

b)

Classification

c)

Linear Regression

d)

Dimensionality Reduction

38.

What does a Linear Regression model fit to the data set?

a)

A complex curve

b)

A line (or hyper-dimensional line)

c)

A series of points

d)

A scattered plot

39.

What is the mathematical equation for a simple linear model?

a)

Y = X^2

b)

Y = a*X + b

c)

Y = log(X)

d)

Y = sin(X)

40.

How does a Polynomial Regression Model handle non-linear relationships?

a)

By ignoring them

b)

By modeling them using a linear model

c)

By converting them to linear relationships

d)

By making them perfectly straight lines

41.

What type of curve does a Polynomial Regression Model fit between X and Y values?

a)

A straight line

b)

A non-linear curve

c)

A dashed line

d)

A horizontal line

42.

What are Decision Tree Regression models sensitive to?

a)

Inliers

b)

Missing values

c)

Outliers

d)

Balanced data

43.

What does a Support Vector Regression model build to cover most data points?

a)

A single point

b)

A hyperplane with a maximum margin

c)

A small circle

d)

A random boundary

44.

What technique do Ridge/Lasso regression models use to reduce model complexity?

a)

Simplification

b)

Regularization

c)

Expansion

d)

Normalization

45.

What is the "small bias" introduced by Ridge/Lasso regression models known as?

a)

Bonus

b)

Reward

c)

Penalty

d)

Offset

46.

Which error metric calculates the absolute difference between actual and predicted values?

a)

Mean Squared Error (MSE)

b)

Root Mean Square Error (RMSE)

c)

Mean Absolute Error (MAE)

d)

R Squared (R2)

47.

Why is the difference squared in Mean Squared Error (MSE)?

a)

To make values larger

b)

To simplify calculations

c)

To avoid the cancellation of negative terms

d)

To highlight negative errors

48.

What is Root Mean Squared Error (RMSE) a simple square root of?

a)

Mean Absolute Error (MAE)

b)

Mean Squared Error (MSE)

c)

R Squared (R2)

d)

Mean Absolute Percentage Error (MAPE)

49.

What is the possible range for Mean Absolute Percentage Error (MAPE)?

a)

0 to Infinity

b)

Less than 0%

c)

Greater than 100%

d)

Only exactly 100%

50.

What does R Squared (R2) tell about the model?

a)

Its loss value

b)

Its performance

c)

Its complexity

d)

Its computational time