Data Science and Machine Learning (Theory and Projects) A to Z - Classical CNNs: VGG

The video tutorial discusses the VGG convolutional neural network, introduced in 2014, which consists of VGG blocks. Each block typically includes several convolutional layers with 3x3 filters, followed by a Max pooling layer. The VGG network architecture builds on these blocks, with variations in the number of convolutional layers per block. The original VGG network, known as VGG 11, has five blocks with a total of eight convolutional layers and three fully connected layers. The video also mentions other successful VGG models like VGG 16 and VGG 19, and highlights the influence of VGG blocks on later networks like ResNet and Inception.

What is the main architecture introduced in 2014 that is mentioned in the text?

Describe the structure of a VGG block as explained in the text.

What is the purpose of the Max pooling layer in a VGG block?

How many convolutional layers are typically found in the first two VGG blocks?

Explain how the number of output channels changes across the VGG blocks.

What are the nonlinearities used in the VGG network?

How does the spatial dimension of the input tensor change from one VGG block to the next?