Which algorithm serves as a non-parametric approach for supervised learning?

K-nearest Neighbors (KNN) is a non-parametric algorithm that is widely used for supervised learning tasks, such as classification and regression. The term "non-parametric" in this context means that KNN does not assume a predefined form for the underlying data distribution; instead, it makes decisions based on the traits of the training data itself.

In KNN, the algorithm finds the 'k' closest data points (neighbors) to a query point and makes predictions based on the majority class (for classification) or the average (for regression) of those neighbors. This approach is inherently flexible, as KNN can adapt to the structure of the data without being restricted by an assumed model form.

This characteristic differentiates KNN from parametric methods, such as linear regression, which require specific assumptions about the data distribution and involve the estimation of parameters. Support Vector Machines and Decision Trees, while capable of capturing complex patterns, are also considered in a different context regarding their operational characteristics and structure. KNN's reliance on direct computation of distances in the input space further emphasizes its non-parametric nature.