模式识别经典教材.pdfVIP

Lecture 8: The K Nearest Neighbor Rule (k-NNR) g Introduction g k-NNR in action g k-NNR as a lazy algorithm g Characteristics of the k-NNR classifier g Optimizing storage requirements g Feature weighting g Improving the nearest neighbor search Introduction to Pattern Recognition 1 Ricardo Gutierrez-Osuna Wright State University Introduction g The K Nearest Neighbor Rule (k-NNR) is a very intuitive method that classifies unlabeled examples based on their similarity with examples in the training set n For a given unlabeled example x ∈ℜD, find the k “closest” labeled examples u in the training data set and assign xu to the class that appears most frequently within the k-subset g The k-NNR only requires n An integer k n A set of labeled examples (training data) n A metric to measure “closeness” ω 1 ω g Example 2 n In the example below we have three classes x and the goal is to find a class label for the u unknown example xu n In this case we use the Euclidean distance and a value of k=5 neighbors ω n Of the 5 closest neighbors, 4 belong to ω1 and 3 1 belongs to ω , so xu is assigned to ω , the 3 1 predominant class