wayground logo

Free Printable Worksheets

NEW

Font size

S
M
L
XL
Worksheets

Content Section | Worksheet Questions (Grade 13)

Total questions: 20

Worksheet time: 4mins

Name
Class
Date
1.

What is the main purpose of data augmentation?

a)

Reduce training time

b)

Increase dataset size artificially

c)

Remove noise from data

d)

Reduce model parameters

2.

Data augmentation is most commonly used in which domain?

a)

Text processing

b)

Image processing

c)

Time series analysis

d)

Reinforcement learning

3.

Early stopping is used to prevent:

a)

Underfitting

b)

High bias

c)

Overfitting

d)

Gradient explosion

4.

L1 regularization adds which penalty term to the loss function?

a)

Sum of squared weights

b)

Sum of absolute weights

c)

Maximum weight value

d)

Mean of weights

5.

L2 regularization penalizes:

a)

Absolute value of weights

b)

Number of features

c)

Squared value of weights

d)

Dropout probability

6.

Dropout works by:

a)

Increasing neurons

b)

Removing layers

c)

Randomly disabling neurons during training

d)

Normalizing inputs

7.

Dropout is applied during:

a)

Inference only

b)

Training only

c)

Validation only

d)

Both training and inference

8.

Batch normalization normalizes:

a)

Weights

b)

Loss values

c)

Activations of a layer

d)

Labels

9.

Batch normalization is usually applied:

a)

After activation

b)

Before activation

c)

At output layer only

d)

After loss computation

10.

Which technique randomly drops neurons to reduce overfitting?

a)

L2 Regularization

b)

Batch Normalization

c)

Dropout

d)

Early Stopping

11.

How does data augmentation help generalization?

a)

Increasing model complexity

b)

Exposing model to data variations

c)

Reducing training epochs

d)

Increasing batch size

12.

Which statement correctly compares L1 and L2 regularization?

a)

Both produce sparse weights

b)

L1 produces sparse weights, L2 shrinks weights

c)

L2 sparse, L1 shrinks weights

d)

Both remove weights

13.

Why is dropout turned off during inference?

a)

Speed up computation

b)

Use full network capacity

c)

Reduce memory usage

d)

Avoid normalization

14.

One benefit of batch normalization is:

a)

Increased overfitting

b)

Slower convergence

c)

Reduced sensitivity to initialization

d)

Larger gradients

15.

Early stopping typically monitors:

a)

Training accuracy

b)

Validation loss

c)

Learning rate

d)

Weight decay

16.

Why does L1 regularization lead to sparse solutions?

a)

Squares weights

b)

Uniform scaling

c)

Sharp penalty at zero

d)

Increases learning rate

17.

L2 regularization affects weight updates by:

a)

Increasing magnitude

b)

Pushing weights toward zero

c)

Removing weights randomly

d)

Normalizing gradients

18.

Batch normalization allows higher learning rates because it:

a)

Reduces model size

b)

Stabilizes input distributions of layers

c)

Increases batch size

d)

Removes regularization

19.

Why are dropout and batch normalization sometimes not used together?

a)

Increase training time

b)

Dropout disrupts batch statistics

c)

BN removes dropout effect

d)

Both reduce capacity

20.

When can data augmentation hurt performance?

a)

Distorts true data distribution

b)

Large batch size

c)

L2 regularization used

d)

Early stopping applied