A startup has a small dataset of 5,000 labeled images of industrial defects. They want to build a classifier using a pretrained ResNet50. They have limited GPU time and need to train quickly while avoiding overfitting. Which approach is most appropriate?

Freeze all convolutional layers and train only the new fully connected head

Replace the final fully connected layer with a larger one and train only that layer

Randomly initialize all layers and train the entire network from scratch

Fine-tune all layers with a very high learning rate

What is the primary function of a max pooling layer in a CNN?

To reduce the spatial dimensions of the feature maps while retaining the most prominent features.

To increase the spatial resolution of the feature maps for better detail.

To apply a learnable nonlinear transformation to each pixel.

To normalize the activations across the channel dimension.

A deep CNN is training with batch normalization and ReLU activations. The training loss is decreasing, but the validation loss is increasing. The team suspects overfitting. They want to add regularization without significantly increasing training time. Which technique is most appropriate?

Add dropout layers after the convolutional layers.

Remove batch normalization layers.

GIAC Machine Learning Engineer

Domain 3Objective 2

Convolutional Neural Networks GMLE Practice Questions (Page 7)

Part of the Advanced Machine Learning and Neural Networks domain, which makes up ~30% of our current practice bank. GIAC (SANS) does not publish an official question count, but from its 180-minute exam (~70–120 total, ~21–36 in this domain), expect 7–12 from this objective — we provide 40 practice questions to prepare you well beyond it. (estimate)

40questions here

8free pages

10concepts

Questions 31–35

31application · medium
A startup has a small dataset of 5,000 labeled images of industrial defects. They want to build a classifier using a pretrained ResNet50. They have limited GPU time and need to train quickly while avoiding overfitting. Which approach is most appropriate?
Select an answer first
32foundation · easy
What is the primary function of a max pooling layer in a CNN?
Select an answer first
33foundation · easy
For a task that requires assigning a class label to every pixel in an image, such as medical image segmentation, which type of CNN architecture is most appropriate?
Select an answer first
34expert · hard
A deep CNN is training with batch normalization and ReLU activations. The training loss is decreasing, but the validation loss is increasing. The team suspects overfitting. They want to add regularization without significantly increasing training time. Which technique is most appropriate?
Select an answer first
35foundation · easy
Which CNN architecture is specifically designed to output bounding boxes and class labels for multiple objects in an image?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to GMLE

Free Basic Practice is a study aid with revealable answers — not a scored exam. Examers.io is independent and not affiliated with or endorsed by GIAC (SANS). “GMLE” is a trademark of its owner, used for identification only.

Convolutional Neural Networks GMLE Practice Questions (Page 7)

Questions 31–35

A startup has a small dataset of 5,000 labeled images of industrial defects. They want to build a classifier using a pretrained ResNet50. They have limited GPU time and need to train quickly while avoiding overfitting. Which approach is most appropriate?

What is the primary function of a max pooling layer in a CNN?

For a task that requires assigning a class label to every pixel in an image, such as medical image segmentation, which type of CNN architecture is most appropriate?

A deep CNN is training with batch normalization and ReLU activations. The training loss is decreasing, but the validation loss is increasing. The team suspects overfitting. They want to add regularization without significantly increasing training time. Which technique is most appropriate?

Which CNN architecture is specifically designed to output bounding boxes and class labels for multiple objects in an image?

Keep studying — Advanced Machine Learning and Neural Networks