AI Paper Atlas

Research map / Learning theory and algorithms / Sample complexity

Smoothing Structured Decomposable Circuits

Andy Shih, Guy Van den Broeck, Paul Beame, Antoine Amarilli

NeurIPS 2019 · Learning theory and algorithms › Sample complexity · approach: Computational thresholds in graphs and structure · 2 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use the near-linear structured-decomposable smoothing algorithm instead of naive quadratic smoothingHand-crafted circuits · Hand-crafted circuits, sizes 40k–8,500ksmoothing time: 0.82 ± 0.01 -> 0.04 ± 0.01; 50 ± 0.3 -> 0.31 ± 0.01; 293 ± 2 -> 0.74 ± 0.04; 6050 ± 20 -> 4.13 ± 0.09Improved
Replace the original All-Marginals subroutine with the linear-time algorithmCollapsed sampling on Segmentation-11 · Segmentation-11, sizes 100k–400kAll-Marginals operations: 28,494 ± 598 -> 20,207 ± 411; 55,875 ± 1,198 -> 36,101 ± 1,522; 86,886 ± 6,330 -> 56,094 ± 817Improved
Replace the original All-Marginals subroutine with the linear-time algorithmCollapsed sampling on DBN-11 · DBN-11, sizes 100k–400kAll-Marginals operations: 172,610 ± 1,821 -> 26,807 ± 644; 344,748 ± 3,881 -> 51,864 ± 851; 626,235 ± 9,985 -> 99,567 ± 697Improved
Replace the original All-Marginals subroutine with the linear-time algorithmCollapsed sampling on CSP-13 · CSP-13, sizes 100k–400kAll-Marginals operations: 36,531 ± 1,484 -> 20,814 ± 619; 90,352 ± 3,593 -> 38,670 ± 1,438; 122,208 ± 9,971 -> 55,269 ± 1,819Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy