AI Paper Atlas

Research map / Probabilistic and scientific modeling / Molecules and proteins

Learning Graph Models for Retrosynthesis Prediction

NeurIPS 2021 · Probabilistic and scientific modeling › Molecules and proteins · approach: Graph models for retrosynthesis · 71 citations (OpenAlex)

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's title only (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 MPN-based edit score updates over bond dependency graph (ablation)GraphRetro edit prediction · USPTO-50k, class unknowntop-1 accuracy (%): 70.1 -> 70.8Helped
add MPN-based edit score updates (ablation)GraphRetro edit prediction · USPTO-50k, class knowntop-1 accuracy (%): 84.3 -> 84.6No clear effect
predict only on existing bonds and atoms instead of all atom pairs (ablation)GraphRetro edit prediction · USPTO-50k, class knowntop-1 accuracy (%): 81.9 -> 84.6Helped
predict only on existing bonds and atoms instead of all atom pairs (ablation)GraphRetro edit prediction · USPTO-50k, class unknowntop-1 accuracy (%): 68.6 -> 70.8Helped
shared MPN for leaving group representations instead of independent embeddings (ablation)synthon completion · USPTO-50k, class knowntop-1 accuracy (%): 77.4 -> 76.9No clear effect
leaving group vocabulary completion and edit-based semi-template design vs previous semi-template methodssemi-template retrosynthesis · USPTO-50k, class unknowntop-1 accuracy (%): 50.4 -> 53.7Helped

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