So far, we've learned how convolutional and pooling layers work. A CNN can have multiple convolutional layers and pooling layers. However, these layers will only extract features from the input image and produce the feature map; that is, they are just the feature extractors.

Given any image, convolutional layers extract features from the image and produce a feature map. Now, we need to classify these extracted features. So, we need an algorithm that can classify these extracted features and tell us whether the extracted features are the features of a horse, or something else. In order to make this classification, we use a feedforward neural network. We flatten the feature map and convert it into a vector, and feed it as an input to the feedforward network. The feedforward network takes this flattened feature map as an input, applies an activation function, such as sigmoid, and returns the output, stating whether the image contains a horse or not; this is called a fully connected layer and is shown in the following diagram: