An Improved Diverse Density Algorithm for Multiple Overlapped Instances
Lei Xu, Maozu Guo, Quan Zou, Yang Liu, Haifeng Li · 2008
Multiple-instance learning is a special machine learning algorithm between supervised learning and unsupervised learning, which has been used in medicine design, image retrieval and other research fields, and attained good performance. Diverse Density (DD) algorithm is a typical multiple- instance learning method. Due to the character of sparse positive instances, when classifying the bags which include multiple overlapped instances, some negative bags are considered as positive bags. To solve this problem, this paper proposed a new classification method, which modifies the influence strategy of the instances to the bags when classifying the bags. To verify the method, it is used to classify the real and pseudo microRNA precursors in bioinformatics, and has obtained exciting results.