arXiv · cs.AI · Ziyi Guan, Yael Tauman Kalai, Ron Rothblum, Zoe Xi · 9 小时前模型评分43
可回退交互证明（RIPs）：用回退机制实现高效可扩展监督
Efficient Scalable Oversight via Rewinding
导读
研究者提出可回退交互证明（RIPs，Rewindable Interactive Proofs），让弱验证者在交互中把强 人工智能（AI，Artificial Intelligence） 证明者回退到早前状态再继续验证，无需假设证明者诚实，也不使用密码学。该方案中诚实证明者运行时间为原计算的线性量级，验证者一次性预处理后接近输入长度的线性时间，初步 MNIST 实验显示可快于现有部署的证明系统。
分类置信度
中 · 待人工复核
- 只有单一媒体来源，需要人工复核。
来源级别：媒体与个人
事件类型：研究发布
证据摘录
- “In this work, we introduce \emph{rewindable interactive proofs} (RIPs), a new model of relativizing interactive proofs for scalable oversight.”
- “In our construction the honest prover runs in time linear in that of the original computation, and the verifier, after a one-time preprocessing step, runs in time nearly linear in the input length (which can be much smaller than the computation itself).”
- “We implement our protocol, report a preliminary MNIST experiment, and show that an RIP can be faster than a state-of-the-art deployed proof system.”
来源：arXiv · cs.AI · arxiv.org