A convolutional neural network, or CNN, is a deep learning model designed to find local patterns in data. It is best known for image tasks because its layers scan small regions of an input and learn features such as edges, textures, and shapes. That makes CNNs efficient at spotting repeated visual structures without needing hand-written rules for every pattern.
In cyber security, CNNs are useful when malware, memory dumps, or network traffic are converted into image-like representations. The model can then classify suspicious samples, group related threats, or help analysts prioritize what to inspect. Attackers benefit from the same idea when they try to hide malicious behavior inside complex binaries, so defenders use CNNs to detect subtle structure that traditional signatures may miss. Their value depends on good training data and careful testing, because a CNN trained on one threat family may not generalize well to new packing methods or changing attack styles.