K-Nearest Neighbors (KNN) is a simple, supervised machine learning algorithm used for classification and regression. It works by finding the k-nearest neighbors to a given data point and making predictions based on the majority vote or average of these neighbors.