Deep Learning CNN Convolutional Neural Networks with Python - CNN Example

The video tutorial explains the construction of a simple convolutional neural network (CNN) for grayscale images. It covers the setup of convolutional layers with filters, the application of ReLU nonlinearity, and the use of max pooling. The tutorial also discusses flattening the output for a fully connected layer with a softmax function and defines a squared loss function. Finally, it introduces backpropagation and weight updates, emphasizing the potential to extend these concepts to larger networks.

What is the size of the input image used in the CNN example?

• 64 by 64

• 32 by 32

• 16 by 16

• 28 by 28

How many convolutional filters are applied to the input image?

• Four

• One

• Two

• Three

What nonlinearity is applied after adding the bias in the convolution process?

• ReLU

• Sigmoid

• Tanh

• Softmax

What is the purpose of max pooling in the CNN process?

• To increase the size of the feature map

• To reduce the dimensionality of the feature map

• To apply nonlinearity

• To add bias

What is the dimension of the pooled feature map after max pooling?

• 4 by 4

• 16 by 16

• 8 by 8

• 32 by 32

What function is used in the fully connected layer to classify the output?

• Softmax

• Sigmoid

• ReLU

• Tanh

What type of loss function is used in this CNN example?

• Hinge loss

• Squared loss

• Cross-entropy loss

• Log loss