AI Paper Atlas

Research map / Graph learning / Graph algorithms and clustering

Efficient Learning of Discrete-Continuous Computation Graphs

David Friede, Mathias Niepert

NeurIPS 2021 · Graph learning › Graph algorithms and clustering · approach: Neural combinatorial optimization on graphs

This paper tried a common change: Add dropout

add dropout residual connections (DR) · ListOps · task accuracy: 74.8 -> 96.1 Improved

What 795 accepted papers report for this change

See this change in detailMatched by the wording of each paper's own result, read by AI from the full text. Papers report gains more often than failures, so this shows what gets published, not the chance a change works.

Read the paper

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
add temperature matching (TM) (ablation)dropout-residual Gumbel-softmax GNN · ListOpstask accuracy: 95.1 -> 96.1Improved
add temperature matching (TM) (ablation)same model · ListOpsedge precision: 76.9 -> 82.3Improved
add dropout residual connections (DR) (ablation)Gumbel-softmax GNN · ListOpstask accuracy: 74.8 -> 96.1Improved
add dropout residuals (ablation)same model · ListOpsedge precision: 43.9 -> 82.3Improved
add both DR and TM (ablation)same model · ListOpstask accuracy: 56.2 -> 96.1Improved
discrete-continuous model vs best baseline on task accuracyOurs tau=1 · ListOpstask accuracy: 96.8 -> 96.1No clear effect
discrete-continuous model vs LSTM for extrapolationOurs tau=1 · ListOps depth 8extrapolation accuracy: 83.7 -> 92.6Improved
discrete-continuous model vs LSTM for extrapolationOurs tau=1 · ListOps depth 10extrapolation accuracy: 76.9 -> 86.9Improved
train on ground-truth graph instead of sequenceOurs · ListOpstask accuracy: 96.1 -> 99.8Improved
temperature tau=2 vs tau=1Ours · ListOpsedge precision: 82.3 -> 76.8Hurt
discrete-continuous model vs CoKEOurs tau=4 · WordNet path queryH@10: 67.9 -> 64.3Hurt
discrete-continuous model vs CoKEOurs tau=4 · Freebase path queryH@10: 77.7 -> 68.7Hurt

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