A Large-Scale -Nearest Neighbor Classification Algorithm Based on Neighbor Relationship Preservation

Algorithm 1

NPR-kNN algorithm.

Input: The training set , the test instance , number of subsets .

Output: The predicted label .

1 Initialization: , -nearest neighbor set

2 Divide the set into disjoint using Mini-batch -Means clustering algorithm and get cluster centers set ;

 whiledo

3  Find the set according to the similarity between and each instance of , where 1

4  

5  Update -nearest neighbor set of to generate a new set by comparing the similarity between and each instance of

6  ;

7  ;

 end

8 Obtain the predicted label based on the majority class on

9 Return.