Weakly Supervised Learning in a Group of Learners with Communication

In the initial research, we executed these inferences according to the limited graph that we used as input, although the approximate inference can also be used. We have designed a system where learners with little knowledge of the problem can label unlabeled data with each other’s help and this data can be used to train learners. It should be noted that sharing opinions It causes solidarity. In our proposed method, the accuracy of the system is controlled by preventing the correlation effect between users in voting for labeling unlabeled data.