Traditional classification methods assume that the training and the test data arise from the same underlying distribution. However, in several adversarial settings, the test set is...
Abstract--The automatic discovery of group conversational behavior is a relevant problem in social computing. In this paper, we present an approach to address this problem by defin...