Skip to content
Hot eventLive

Hierarchy-GBP:通过抽象与恢复加速因子图推理

1 reports1 sources3 hr ago updated

Get the story

AI overview

研究者提出 Hierarchy-GBP(H-GBP),一种迭代式两阶段框架:先用粗图近似求解全局误差并投影回原图,再用 GBP 细化局部误差。论文证明 H-GBP 收敛到最优,实验显示其在线性稀疏图上比标准 GBP 收敛更快。该成果发表于 arXiv,归属 Computer Vision 方向。

Generated from reports · updated 3 hr ago

Timeline

Follow the coverage from different angles.

Oct 7, 2026
  1. arXiv · Computer Vision
    Hierarchy-GBP:通过抽象与恢复加速因子图推理

    研究者提出 Hierarchy-GBP(H-GBP),一种迭代式两阶段框架,先用粗图近似求解全局误差并投影回原图,再用 GBP 细化局部误差。论文证明 H-GBP 收敛到最优,并在实验中显示其在线性稀疏图上比标准 GBP 收敛更快。

Heat trend

Current heat 9·Comparable peak 10(Oct 7)·Comparable change over 24 hours –

02.557.510Oct7Oct7Oct7Oct7

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.