試験
通常
試験時間: 00:00問題時間: 00:00
問題一覧
問題 10 / 65
Which approach would be considered effective when there is almost no labeled fraudulent data in the transaction logs?
①Detect unknown anomalous patterns using unsupervised learning and anomaly detection.
②Because there are few labels, AI cannot be used, so we'll restrict ourselves to rule-based approaches.
③It is realistic to always manually scrutinize all transactions.
④If there are no labels, evaluating the model is unnecessary.
⑤We should discard the data and redirect resources to other tasks.