Concept: In deep learning, convolutional neural networks (CNNs) play a central role in processing and classifying images. These models extract important features from images and pass them through various layers to generate predictions. The feature extraction process helps transform a large number of raw pixel values into a more compact and meaningful representation that supports label prediction. CNNs are composed of multiple specialized layers, each contributing to feature extraction or label prediction. Which type of layer is responsible for reducing the number of extracted feature values while preserving the most relevant distinguishing characteristics?