AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Self-Improved Retrosynthetic Planning

Junsu Kim, Sungsoo Ahn, Hankook Lee, Jinwoo Shin

ICML 2021 · Architectures and efficiency › RNNs and normalization · approach: Neural-symbolic reasoning · 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 proposed augmentation instead of none (ablation)Retrosynthetic planning · N=250 single-step model callssuccess rate: 81.05 -> 83.16Improved
Use the proposed augmentation instead of none (ablation)Retrosynthetic planning · N=500 single-step model callssuccess rate: 92.63 -> 93.16Improved
Replace mixed forward/reverse augmentation with the proposed augmentation (ablation)Retrosynthetic planning · N=50 single-step model callssuccess rate: 52.63 -> 50.00Hurt
Replace mixed forward/reverse augmentation with the proposed augmentation (ablation)Retrosynthetic planning · N=250 and N=500 single-step model callssuccess rate: 66.32 -> 83.16; 72.63 -> 93.16Improved
Use a positive threshold instead of Thr=0 (ablation)Single-step model · Single-step modelTop-1: 41.04 -> 44.45Improved
Vary Thr from 0.5 to 0.9 (ablation)Single-step model · Single-step modelTop-10: 73.26–73.30No clear effect
Use a positive threshold instead of Thr=0 (ablation)Retrosynthetic planning · N=500success rate: 83.58 -> 92.21Improved

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