ZeNOVA：流形约束初始噪声优化用于高效生成模型对齐

1. 12:00
   arXiv · Machine Learning Theory

   ZeNOVA：流形约束初始噪声优化实现高效生成模型对齐

   ZeNOVA 提出一种无梯度初始噪声对齐方法，通过退火软值引导、流形约束超球面朗之万动力学和 Metropolis-Hastings 跳跃解决黑盒奖励场景下的不稳定与低效问题。实验在图像和视频生成模型上表明，ZeNOVA 比所有零阶基线更稳定地将初始噪声优化至更高奖励，利用高斯先验几何结构。方法适用于多种黑盒奖励对齐场景。

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.