Research map / Probabilistic and scientific modeling / Molecules and proteins
T2T: From Distribution Learning in Training to Gradient Search in Testing for Combinatorial Optimization
Yang Li, Jinpei Guo, Runzhong Wang, Junchi Yan
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 gradient search (1 iteration) (ablation) | T2T · TSP-500 greedy | performance drop (%): 1.59% -> 0.94% | Improved |
| increase gradient-search iterations from 1 to 3 (ablation) | T2T · TSP-500 greedy | performance drop (%): 0.94% -> 0.80% | Improved |
| increase gradient-search iterations from 3 to 10 (ablation) | T2T · TSP-500 greedy | performance drop (%): 0.80% -> 0.72% | Improved |
| increase graph sparsification k from 10 to 200 (ablation) | T2T · TSP-500 greedy | performance drop (%): 0.83% -> 0.74% | Improved |
| vary disruption ratio alpha (ablation) | T2T · TSP-500 greedy and sampling | performance drop (%) | Mixed |
| vary number of denoising steps for initial solution (ablation) | T2T · TSP-500 | performance drop (%) | No clear effect |
| add objective-guided gradient search to DIFUSCO | DIFUSCO · TSP-50 greedy | performance drop (%): 0.48% -> 0.04% | Improved |
| add objective-guided gradient search to DIFUSCO | DIFUSCO · TSP-100 greedy | performance drop (%): 1.01% -> 0.18% | Improved |
| add gradient search to DIFUSCO | DIFUSCO · TSP-500 greedy | performance drop (%): 9.41% -> 5.09% | Improved |
| add gradient search to DIFUSCO | DIFUSCO · TSP-500 sampling+2OPT | performance drop (%): 0.83% -> 0.37% | Improved |
| add gradient search to DIFUSCO | DIFUSCO · TSP-1000 sampling+2OPT | performance drop (%): 1.30% -> 0.78% | Improved |
| add gradient search to DIFUSCO | DIFUSCO · TSPLIB | performance drop (%): 0.319% -> 0.133% | Improved |
Closest papers by meaning
- Fast T2T: Optimization Consistency Speeds Up Diffusion-Based Training-to-Testing Solving for Combinatorial Optimization (NeurIPS 2024 · similarity 0.94)
- Position: Rethinking Post-Hoc Search-Based Neural Approaches for Solving Large-Scale Traveling Salesman Problems (ICML 2024 · similarity 0.89)
- Neural Solver Selection for Combinatorial Optimization (ICML 2025 · similarity 0.89)
- Improving Generalization of Neural Combinatorial Optimization for Vehicle Routing Problems via Test-Time Projection Learning (NeurIPS 2025 · similarity 0.88)
- PolyNet: Learning Diverse Solution Strategies for Neural Combinatorial Optimization (ICLR 2025 · similarity 0.87)
- Complexity Scaling Laws for Neural Models using Combinatorial Optimization (NeurIPS 2025 · similarity 0.86)
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.
