Active Learning Based on Transfer Learning Techniques for Text Classification

Text preprocessing is a common task in machine learning applications that involves hand-labeling sets. Although automatic and semi-automatic annotation of text data growing field, researchers need to develop models use resources as efficiently possible for task. The goal this work was learn faster with fewer resources. In paper, the combination active transfer examined purpose developing an effective categorization method. These two forms have proven their efficiency capacity train correct substantially less training data. We considered three types criteria selecting points: random selection, uncertainty sampling criterion selection. Experimental evaluation performed on five sets from different domains. findings experiments suggest by combining learning, algorithm performs better labels than selection points.