Nous：学习与认证记忆决策优于源校准

1. 12:00
   arXiv · Machine Learning Theory

   Nous：学习与认证记忆决策优于源校准

   Nous 提出在源校准前学习与认证记忆决策的理论框架，在四模型隐马尔可夫族上证明学习与认证改进所需样本量为 Θ(l^-2)，而固定精度源估计需 Θ(l^-4)，决策可少用平方级样本。该方法在 MiniGrid 的 45,000 条可变状态历史和 9,000 个回合中测试，新证书接受 9/9 改进和 4/9 优于 last-write-wins，而旧证书为 0。

Current heat 9·Comparable peak 10（Oct 2）·Comparable change over 24 hours –

The trend compares only the same participants observed continuously; its range may be smaller than the current heat count. Move or click on the chart to inspect hourly heat; use the left and right arrow keys to switch.