AI Paper Atlas

Research map / Graph learning / Knowledge and temporal graphs

Versatile Multi-stage Graph Neural Network for Circuit Representation

Shuwen Yang, Zhihao Yang, Dong Li, Yingxue Zhang, Zhanguang Zhang, Guojie Song, Jianye HAO

NeurIPS 2022 · Graph learning › Knowledge and temporal graphs · approach: Heterogeneous graph learning · 5 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
remove topo-edges from Circuit Graph (ablation)Circuit GNN · ISPD2011 congestion, placementgrid pearson: 0.697 -> 0.684Hurt
remove topo-edges from Circuit Graph (ablation)Circuit GNN · ISPD2011 congestion, placementcell pearson: 0.887 -> 0.883Hurt
remove topo-edges from Circuit Graph (ablation)Circuit GNN · ISPD2011 congestion, placementgrid spearman: 0.770 -> 0.730Hurt
remove topo-edges from Circuit Graph (ablation)Circuit GNN · ISPD2011 congestion, placementgrid kendall: 0.577 -> 0.536Hurt
add geometric edges and cell positions in placement stage vs topology onlyCircuit GNN · ISPD2011 congestion, logic synthesis vs placementcell pearson: 0.779 -> 0.887Improved
use Circuit GNN instead of CongestionNetCircuit GNN · ISPD2011 congestion, logic synthesisgrid pearson: 0.277 -> 0.315Improved
use Circuit GNN instead of MPNN (5x faster)Circuit GNN · ISPD2011 congestion, logic synthesisgrid pearson: 0.292 -> 0.315Improved
use Circuit GNN instead of LHNNCircuit GNN · ISPD2011 congestion, placementgrid pearson: 0.703 -> 0.697Mixed
use Circuit GNN instead of LHNNCircuit GNN · DAC2012 net wirelength, placementMAE: 0.581 -> 0.483Improved
use Circuit GNN instead of LHNNCircuit GNN · DAC2012 net wirelength, placementpearson: 0.801 -> 0.848Improved
use Circuit GNN features frozen (evaluate) instead of LHNNCircuit GNN · Transfer congestion to wirelengthpearson: 0.689 -> 0.799Improved
fine-tune GNN instead of freezingCircuit GNN · Transfer congestion to wirelengthpearson: 0.799 -> 0.842Improved

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