試験
通常
試験時間: 00:00問題時間: 00:00
問題一覧
問題 22 / 65
Which data collection strategy is most effective for efficiently increasing labeled rare disease data when labeling costs are high?
①Randomly select samples and label them.
②Have experts label all data evenly.
③Learn features via self-supervised learning, then label all samples at once.
④Use active learning that prioritizes labeling samples with high model uncertainty.
⑤Mechanically duplicate data to increase the number of existing labels.