Research Article
Iterative Nearest Neighborhood Oversampling in Semisupervised Learning from Imbalanced Data
Algorithm 1
Iterative Nearest Neighborhood Oversampling (INNO).
Input: NN graph, affinity matrix , stop parameter , imbalanced labeled dataset and | unlabeled dataset ; | Output: balanced or approximate balanced labeled dataset. | Procedure: | 1 while | 2 Initialization , , ; | 3 for each labeled sample in class | 4 for each neighbors of | 5 skip the if it is in or has edges between labeled samples in other class; | 6 if , then update max, | 7 end for | 8 end for | 9 if // all the neighbors of labeled samples in class have edges with labeled samples | in other classes, then , continue; | 10 label with class , remove it from , add it to , ; | 11 end while |
|