摘要：

In many real applications, there are plentiful unlabeled data but limited labeled data, and the acquisition of labels is costly. Active learning reduces the labeling cost by iteratively selecting the most valuable data to query their labels from the annotator. This talk will introduce our recent efforts on cost-sensitive active learning, where the cost varies over different examples, labels and oracles.

在许多实际应用中，有大量的未标记数据，但标记数据有限，获取标签的成本很高。主动学习通过迭代选择最有价值的数据从注释器中查询它们的标签来降低标签成本。本次演讲将介绍我们最近在成本敏感型主动学习方面所做的工作，其中成本因不同的示例、标签和Oracle而异。

query 疑问讯问；oracle 预言