Research map / Image generation / Fast diffusion sampling
Boosting Cross-problem Generalization in Diffusion-Based Neural Combinatorial Solver via Inference Time Adaptation
Haoyu Lei, Kaiwen Zhou, Yinchuan Li, Zhitang Chen, Farzan Farnia
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 DIFU-Ada inference-time adaptation to TSP-trained DIFUSCO | DIFUSCO · PCTSP-20 zero-shot | optimality gap: 19.21% -> 4.20% | Improved |
| add DIFU-Ada inference-time adaptation to TSP-trained DIFUSCO | DIFUSCO · PCTSP-100 zero-shot | optimality gap: 32.31% -> 9.61% | Improved |
| add DIFU-Ada to TSP-trained DIFUSCO | DIFUSCO · OP-100 zero-shot | optimality gap: 20.02% -> 8.06% | Improved |
| apply DIFU-Ada on top of T2T instead of plain T2T | T2T · PCTSP-50 zero-shot | optimality gap: 9.38% -> 3.57% | Improved |
| apply DIFU-Ada compared with plain DIFUSCO | DIFUSCO · PCTSP-20 inference time | time per instance: 1.04s -> 1.72s | Hurt |
| increase number of recursive renoising-denoising iterations (ablation) | DIFU-Ada · PCTSP and OP | optimality gap | Improved |
| add energy-guided sampling during recursive iterations (ablation) | DIFU-Ada · PCTSP and OP | optimality gap | Improved |
| raise guidance temperature (ablation) | DIFU-Ada · PCTSP and OP | optimality gap | Mixed |
| vary penalty coefficient mu in energy function (ablation) | DIFU-Ada · PCTSP and OP | optimality gap | No clear effect |
| use few renoising steps and single denoising step per iteration instead of full recursion (ablation) | DIFU-Ada · inference cost | inference speed: 5-10x speedup | Improved |
Closest papers by meaning
- Fast T2T: Optimization Consistency Speeds Up Diffusion-Based Training-to-Testing Solving for Combinatorial Optimization (NeurIPS 2024 · similarity 0.90)
- Neural Solver Selection for Combinatorial Optimization (ICML 2025 · similarity 0.88)
- A Diffusion Model Framework for Unsupervised Neural Combinatorial Optimization (ICML 2024 · similarity 0.88)
- Efficient Combinatorial Optimization via Heat Diffusion (NeurIPS 2024 · similarity 0.88)
- Improving Generalization of Neural Combinatorial Optimization for Vehicle Routing Problems via Test-Time Projection Learning (NeurIPS 2025 · similarity 0.87)
- Learning to Insert for Constructive Neural Vehicle Routing Solver (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.
