Dynamic Fuzzy Semisupervised Multitask Learning

Semi supervised multitask learning has been one of the hotest problems in the machine learning field in recent years. In this paper a dynamic fuzzy semi supervised multitask learning algorithm has been proposed to deal with the dynamic fuzzy problems. Our goal is to learn dynamic fuzzy possibility of each class with a limited initial data set, and the dynamic fuzzy possibility is numerically equal to the membership functions of each class. So the dynamic fuzzy possibility needs to be adapted with new data incoming. Experiment results shows that our method has performed much better, compared with the semi supervised fuzzy pattern matching algorithm proposed in [9].