Research map / Graph learning / Graph algorithms and clustering
Generalized Schrödinger Bridge on Graphs
Panagiotis Theodoropoulos, Juno Nam, Evangelos Theodorou, Jaemoo Choi
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 |
|---|---|---|---|
| add free-energy running cost f (ablation) | GSBoG · Chignolin folding | Energy barrier (kBT): 1.80 -> 1.39 | Improved |
| add free-energy running cost f (ablation) | GSBoG · Chignolin folding | Fold rate (%): 99.70 -> 99.36 | No clear effect |
| add congestion running cost f (ablation) | GSBoG · Supply chain | Peak node occupancy: nearly halves the peak occupancy | Improved |
| use GSBoG instead of W1 flow | W1 flow · Supply chain | Mean congestion (top-100 nodes): 178.19 -> 21.13 | Improved |
| use GSBoG instead of Attr. Flow | Attr. Flow · Supply chain | Total variation: 0.27 -> 0.03 | Improved |
| use GSBoG instead of Attr. Flow | Attr. Flow · Supply chain | Max edge flow / capacity: 1.59 -> 0.64 | Improved |
| use GSBoG instead of GrSB | GrSB · Chignolin folding | Fold rate (%): 25.11 -> 99.36 | Improved |
| use GSBoG instead of committor-guided bias | Committor-guided bias · Chignolin folding | Energy barrier (kBT): 6.59 -> 1.39 | Improved |
| increase assignment size n from 10 to 20 (ablation) | GSBoG · Assignment task | Assignment accuracy (%): 100% -> 90% | Hurt |
| increase assignment size n from 6 to 20 (ablation) | GSBoG · Assignment task | Mass on optimal selection (%): 97.3% -> 85.0% | Hurt |
| scale graph to 5x10^4 nodes: GSBoG vs Dynamic OT | Dynamic OT · Scalability | Memory (GB): 22.0 -> 1.4 | Improved |
| increase rollout budget from 10^3 to 4x10^3 (ablation) | GSBoG · Scalability | Runtime: 10^3 -> 4x10^3 | Mixed |
Closest papers by meaning
- Generalized Schrödinger Bridge Matching (ICLR 2024 · similarity 0.87)
- Discrete Diffusion Schrödinger Bridge Matching for Graph Transformation (ICLR 2025 · similarity 0.86)
- Momentum Multi-Marginal Schrödinger Bridge Matching (NeurIPS 2025 · similarity 0.86)
- Beyond Self-Repellent Kernels: History-Driven Target Towards Efficient Nonlinear MCMC on General Graphs (ICML 2025 · similarity 0.85)
- Trajectory Inference with Smooth Schrödinger Bridges (ICML 2025 · similarity 0.84)
- Graph-Based Stochastic-Power-UCT: Monte-Carlo Graph Search with Power Mean Estimation (NeurIPS 2026 · 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.
