DL & RL Module-2 Quiz

Which of the following inspired the development of Convolutional Neural Networks (CNNs)?

• Human skeletal system

• Biological visual cortex

• Human ear

• Digestive system

The main idea of CNNs is to extract __________ at lower layers and combine them into complex patterns at higher layers.

• Global features

• Random weights

• Local features

• Output probabilities

What was the main task of the LeNet architecture?

• ImageNet classification

• Handwritten digit recognition (MNIST)

• Object detection in videos

• Sentiment analysis

Which of the following was a major limitation of LeNet?

• Too many layers

• Wrong non-linearity (sigmoid/tanh)

• Overfitting due to large dataset

• Lack of pooling layers

The vanishing gradient problem occurs when:

• Gradients grow very large

• Gradients disappear in deep networks

• Weights are initialized randomly

• Batch size is too large

Which activation function was introduced in AlexNet to overcome the vanishing gradient problem?

• Sigmoid

• Tanh

• ReLU

• Softmax

Which hardware made training AlexNet practical?

• CPUs

• TPUs

• GPUs (NVIDIA GTX 580)

• FPGAs

In CNNs, stride refers to:

• The filter size

• The step size by which the filter moves across the input

• The number of channels

• The pooling operation

Which architecture reduced the first convolution filter size from 11×11 to 7×7?

• AlexNet

• ZFNet

• VGGNet

• GoogLeNet

Which CNN architecture used 3×3 filters and increased depth up to 19 layers?

• LeNet

• AlexNet

• VGGNet

• ResNet