Unverified50% confidenceFactExact time
Under decoupled dynamics, coupling among agents is manifested only through the reward function (submodular reward), not through state evolution.
1
Sources
50%
Confidence
Long-term
Relevance
11/4/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedNeuroevolution does not rely on a differentiable loss function and is suitable for sparse-reward scenarios in reinforcement learning.71% similarUnverifiedThis non-stationarity does not come from the environment itself, but is mediated by interactions among agents.69% similarUnverifiedReward functions designed by humans can often only approximately describe true intent, and when a model is powerful enough it will exploit loopholes in the reward function to maximize the numerical reward while violating human intentions.68% similarUnverifiedThe study uses a telescoping-sum argument to show that the cumulative satisfaction return from repeatedly resetting needs does not exceed the theoretical upper bound set by the initial drive, and that the reward mechanism itself does not encourage score-farming behavior.68% similarVerified奖励函数设计不当会导致奖励欺骗（Reward Hacking）现象，即智能体找到在奖励函数上得分高但实际无意义甚至有害的行为路径68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1134363API
curl https://kongchang.com/api/v1/knowledge/claims/1134363MCP
get_claim(id=1134363)