Research map / Graph learning / Graph algorithms and clustering
BDD2Seq: Enabling Scalable Reversible-Circuit Synthesis via Graph-to-Sequence Learning
Mingkai Miao, Jianheng Tang, Guangyu Hu, Hongce Zhang
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 changed | Where | Measured | Effect |
|---|---|---|---|
| replace greedy decoding with Diverse Beam Search (Balance, beam 20) (ablation) | BDD2Seq · 148 circuits | Quantum Cost: 346425 -> 312345 | Improved |
| widen beam from 20 to 50 (Quality mode) (ablation) | BDD2Seq · 148 circuits | Quantum Cost: 312345 -> 307367; time 521.1 -> 2076.2 | Improved |
| increase beam width m from 0 to 20 (ablation) | BDD2Seq Diverse Beam Search · 148 circuits | Quantum Cost: monotonic QC reduction up to m=20; saturates beyond 20 | Improved |
| diversity penalty alpha below 0.2 (ablation) | BDD2Seq (m=20) · 148 circuits | Quantum Cost: insufficient variety for alpha<0.2 | Hurt |
| diversity penalty alpha above 0.3 (ablation) | BDD2Seq (m=20) · 148 circuits | Quantum Cost: low-quality beams for alpha>0.3 | Hurt |
| set diversity penalty alpha to 0.25 (ablation) | BDD2Seq (m=20) · 148 circuits | Quantum Cost: minimum QC at alpha=0.25 | Improved |
| use GAT encoder instead of GCN (ablation) | BDD2Seq · Test set | Kendall's tau: 0.4540 -> 0.6182 | Improved |
| use GAT encoder instead of GraphSAGE (ablation) | BDD2Seq · Test set | Kendall's tau: 0.5978 -> 0.6182 | Improved |
| use BLIF2Graph instead of AIG2Graph (ablation) | BDD2Seq · ISCAS85 test set | Kendall's tau: 0.6038 -> 0.6182 | Improved |
| use BLIF2Graph instead of AIG2Graph (ablation) | BDD2Seq · ISCAS85 | end-to-end wall-clock time: reduced by 32.9% | Improved |
| use BDD2Seq Balance instead of GA | BDD2Seq · 22 benchmark circuits | Quantum Cost: 8454773 -> 135124 | Improved |
| use BDD2Seq Balance instead of GA | BDD2Seq · c880 | Quantum Cost: 35122 (GA) vs 37802 (Balance) | Hurt |
Closest papers by meaning
- A Circuit Domain Generalization Framework for Efficient Logic Synthesis in Chip Design (ICML 2024 · similarity 0.86)
- Training-Free Quantum Architecture Search (AAAI 2024 · similarity 0.85)
- Towards Next-Generation Logic Synthesis: A Scalable Neural Circuit Generation Framework (NeurIPS 2024 · similarity 0.85)
- AI-Powered Algorithm-Centric Quantum Processor Topology Design (AAAI 2025 · similarity 0.85)
- TensorRL-QAS: Reinforcement learning with tensor networks for improved quantum architecture search (NeurIPS 2025 · similarity 0.85)
- QCircuitBench: A Large-Scale Dataset for Benchmarking Quantum Algorithm Design (NeurIPS 2025 · similarity 0.84)
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.
