試験
通常
試験時間: 00:00問題時間: 00:00
問題一覧
問題 45 / 65
I want to iteratively improve a chatbot's behavior using feedback obtained from user interactions. Which learning method is optimal?
①Use reinforcement learning to learn a policy that maximizes reward by treating a dialogue as an episode.
②Use supervised learning to train the utterance–response pairs in the conversation logs on hand just once.
③Unsupervised learning to cluster similar utterances.
④Improving the rule base by manually adding dialogue rules.
⑤A method that simply generates new responses indiscriminately using a generative model.