Combining active and semi-supervised learning for spoken language understanding

摘要

Combined active and semi-supervised learning is used to reduce an amount of manual labeling when training a spoken language understanding model classifier. The classifier may be trained with human-labeled utterance data. Selected data from a group of unselected utterance data may be used for manual labeling via active learning. The classifier may be changed, via semi-supervised learning, based on the data selected from the unselected utterance data.