CNN architecture for image classification in TensorFlow - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is the main purpose of a Convolutional Neural Network (CNN) in image classification?

A CNN automatically learns to detect important features like edges, shapes, and textures from images to classify them correctly.

beginner

What does a convolutional layer do in a CNN?

It applies small filters to the input image to create feature maps that highlight important patterns like edges or textures.

beginner

Why do CNNs use pooling layers?

Pooling layers reduce the size of feature maps, making the model faster and helping it focus on the most important features.

intermediate

What role does the fully connected layer play in a CNN?

It takes the features extracted by convolutional and pooling layers and decides which class the image belongs to.

intermediate

What is the typical output activation function used in CNNs for multi-class image classification?

Softmax activation is used to convert the output scores into probabilities for each class.

What is the first layer usually applied in a CNN for image classification?

A Pooling layer

B Convolutional layer

C Fully connected layer

D Dropout layer

The convolutional layer is the first layer that extracts features from the input image.

Which layer reduces the spatial size of the feature maps in a CNN?

A Convolutional layer

B Fully connected layer

C Pooling layer

D Batch normalization

Pooling layers reduce the size of feature maps to lower computation and focus on important features.

What activation function is commonly used after convolutional layers?

A ReLU

B Tanh

C Softmax

D Sigmoid

ReLU (Rectified Linear Unit) is commonly used to add non-linearity after convolutional layers.

What does the softmax function output represent in image classification?

A Class probabilities

B Binary decision

C Feature map

D Loss value

Softmax outputs probabilities for each class, showing how likely the image belongs to each.

Which layer connects the extracted features to the final classification decision?

A Input layer

B Convolutional layer

C Pooling layer

D Fully connected layer

Fully connected layers use the features to classify the image into one of the classes.

Explain the main components of a CNN architecture used for image classification and their roles.

Think about how the image is processed step-by-step from raw pixels to a class label.

Describe how a CNN processes an input image to produce a classification result.

Follow the flow of data through the CNN layers.