The authors state that learning from non-expert human preferences between pairs of trajectory segments can solve complex RL tasks, including Atari games and simulated robot locomotion, without access to the reward function, and the paper reports feedback on less than 1% of the agent's interactions.
One recorded claim. The claim panel gives its qualification, the source passages with their place in each source, and how it was checked.
What it connects
Deep reinforcement learning from human preferences → states its contribution → The authors state that learning from non-expert human preferences between pairs of trajectory segments can solve complex RL tasks, including Atari games and simulated robot locomotion, without access to the reward function, and the paper reports feedback on less than 1% of the agent's interactions.
Where it appears
No collection page holds this claim. It is listed with its records in the explorer.