Unit 4 - CNN Architecture

What is the primary advantage of using a deeper CNN architecture with more layers?

• Improved generalization and feature extraction
• Reduced overfitting
• Faster training times
• Smaller model size

What is the purpose of the softmax activation function in the output layer of a CNN for image classification?

• To introduce non-linearity
• To reduce the size of the feature maps
• To convert raw scores into class probabilities
• To speed up training

Which technique is commonly used to combat overfitting in CNNs?

• Decreasing the dropout rate
• Increasing the batch size
• Adding more convolutional layers
• Reducing the learning rate

In transfer learning using pre-trained CNN models, what are the typical ways to adapt the model to a new task?

• Fine-tuning the last few layers and adding new fully connected layers
• Keeping all layers frozen and adjusting the learning rate
• Using the pre-trained model without any modifications
• Re-training the entire model from scratch

Which layer type is commonly used to capture high-level features in a CNN?

• Convolutional layer
• Pooling layer
• Fully connected layer
• Activation layer

What is the primary purpose of the ReLU (Rectified Linear Unit) activation function in CNNs?

• To introduce non-linearity
• To normalize the data
• To convert data to binary format
• To reduce the size of feature maps

In a CNN, what does the term "kernel" or "filter" refer to?

• A learning rate for gradient descent
• A type of pooling operation
• A small window used for convolution operations
• The output of a fully connected layer

Which layer in a CNN typically follows the convolutional layers and reduces the spatial dimensions of the feature maps?

• Convolutional layer
• Activation layer
• Pooling layer
• Fully connected layer

What is the primary role of dropout layers in CNNs?

• To increase the size of the feature maps
• To add noise to the input data
• To randomly deactivate neurons during training to reduce overfitting
• To control the learning rate

What is the purpose of the softmax activation function in the output layer of a CNN for multi-class classification?

• To introduce non-linearity
• To increase the size of feature maps
• To convert raw scores into class probabilities
• To prevent overfitting