Concept Overview: At the core of deep learning’s success is a model known as a convolutional neural network (CNN). A CNN extracts features from images and processes them through a fully connected neural network to generate predictions. The feature extraction layers help reduce the number of features from the vast array of pixel values to a compact feature set supporting label prediction. Question: CNNs consist of multiple layers, each serving a distinct purpose in feature extraction or label prediction. Which layer is described as: "One of the most difficult challenges in a CNN is avoiding overfitting, where the model performs well on the training data but fails to generalize to unseen data. To mitigate this, one technique involves including layers that modify feature maps during training. While it might seem counterintuitive, this approach helps prevent the model from becoming overly dependent on specific training images."