試験
通常
試験時間: 00:00問題時間: 00:00
問題一覧
問題 40 / 65
Which is the most appropriate reason why bias remains even if features indicating gender are removed from training?
①Because deletion will inevitably reduce the model's accuracy.
②Because the learning algorithm can recover gender.
③Because there are variables that serve as proxies for gender, bias persists through those proxy variables.
④Because removing it makes overfitting more likely.
⑤Because the size of the training data becomes smaller.