Neural Network Deep Learning Architectures 3

Question 1

What is the key innovation of DenseNet compared to ResNet?

Accepted Answer

Connecting every layer to every subsequent layer

Answer

DenseNet connects each layer to all subsequent layers in a feed-forward fashion, enabling feature reuse and reducing the number of parameters.

Question 2

In the context of deep learning, what is a hypernetwork?

Accepted Answer

A network whose weights are generated by another network

Answer

A hypernetwork is a neural network that generates the weights for another (primary) network, enabling dynamic weight adaptation.

Question 3

What does 'weight sharing' mean in the context of convolutional neural networks?

Accepted Answer

The same filter weights are applied across all spatial positions

Answer

Weight sharing in CNNs means the same convolutional filter (with the same weights) is applied across all spatial locations of the input feature map.

Question 4

Which architecture is most commonly used as the backbone for object detection models like YOLO and Faster R-CNN?

Accepted Answer

Convolutional neural network

Answer

CNN-based architectures (like ResNet, VGG) serve as the feature extraction backbone in most object detection frameworks including YOLO and Faster R-CNN.

Question 5

What is the purpose of the projection head in contrastive self-supervised learning frameworks like SimCLR?

Accepted Answer

To map representations to a space where contrastive loss is applied

Answer

The projection head maps encoder representations to a lower-dimensional space where the contrastive loss is computed, and is discarded after training.

Neural Network Practice Test

Neural Network Practice Test

Neural Network Deep Learning Architectures 3