Sample selection based on multiple incremental decision trees in BSP programming library

The sample selection is a key in the active learning, because it intends to select the best informative sample which has no label from the pool or online. And then the selected sample needs to be added into the training sets for updating the classifier. This paper proposed a new method based on multiple incremental decision trees algorithm to measure the ambiguity of the unlabeled samples for the selection. For accelerating the computing speed, the algorithm is developed in the BSP (Bulk Synchronous Parallel) Programming Library which is a computing model for parallel programming.