A fast partition-based batch-mode active learning technique using SVM classifier

The selection of informative samples known as query selection is the most challenging task in active learning. In this article, a batch-mode active learning technique is presented by defining a novel query function. The proposed technique first divides the unlabeled samples into uniform partitions in one-dimensional feature space according to their distribution in the original feature space. Then to select the most informative samples from the unlabeled pool, one sample from each partition is selected based on an uncertainty criterion defined by exploiting SVM classifier. The number of unlabeled samples selected at each iteration of active learning is determined automatically and depends on the number of non-empty partitions generated. The effectiveness of the proposed technique is measured by comparing it with four state-of-the-art techniques exist in the literature by using four different UCI repository data sets. The experimental analysis proved that the proposed technique is robust and computationally less demanding.