arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.07216cs.LG
收藏

探测神经TSP表示以用于指导性决策支持

Beyond Solving: Prescriptive Probing for Neural Routing Solvers

Foster School of Business, University of Washington, Seattle, WA, USA;Paul G.\ Allen School of Computer Science \& Engineering, University of Washington, Seattle, WA, USA

Reuben Narad, Léonard Boussioux, Michael Wagner

AI总结:

本文研究神经TSP求解器在指导性决策支持中的迁移能力,通过探测器在节点移除和边禁止敏感性任务中取得优于基线的准确率。

摘要:

神经组合优化(NCO)领域训练神经策略解决NP难问题,如旅行商问题(TSP)。我们询问,除了生成好的路线外,训练好的TSP求解器是否学习到可以转移到其他优化相关目标的内部表示,这类似于从其他领域进行迁移学习。我们训练了几种基于注意力的TSP策略,收集其内部激活,并在两个NP难的指导性下游任务上训练探测器,这些任务受现实世界物流场景启发:节点移除敏感性(识别最影响性的节点)和边禁止敏感性(识别最关键的边)。在欧几里得TSP100训练模型上,探测器在两个任务上的表现与现有基线相当。将探测信号与几何特征进行集成,优于最强基线:65%的top-1准确率(vs. 58%基线)对于最佳节点移除任务,73%的top-1准确率(vs. 67%基线)对于最差边识别任务。据我们所知,我们是首次研究神经TSP求解器作为迁移编码器,用于指导性what-if决策支持目标,而不仅仅是路线构建。最后,我们显示转移准确性随着求解器质量在训练和模型规模上增加,表明训练更强的NCO求解器也产生更有用的编码器用于下游目标。我们的代码可在github.com/ReubenNarad/tsp_prescriptive_probe获得。

英文摘要:

Neural combinatorial optimization (NCO) trains fast heuristics for routing problems, but planners often need more than a single solve: they ask which stop to drop, which transition to preserve, or which subset of stops to remove if a route is infeasible. Answering such counterfactual questions by re-solving each candidate is expensive even when the action set is small. We introduce Prescriptive Probing: using the frozen representations of a trained NCO model to rank candidate interventions for what-if decision support. While recent interpretability work descriptively probes what NCO solvers encode, we ask whether those same representations can prescriptively guide intervention choices on real-road routing problems. Where exhaustive labels are available, we train supervised probes from offline re-solve labels; where they are not (e.g., combinatorial action spaces), we train a lightweight reinforcement learning intervention head. Across real-road benchmarks on the Asymmetric Traveling Salesperson Problem and Capacitated Vehicle Routing Problem, probes built on frozen model representations achieve the strongest performance on a majority of intervention tasks and improve over local-repair heuristics on edge-forbiddance and on key node- and multi-node-removal metrics. We further investigate probeability as a property of the routing model itself, varying model quality, the mix of supervised learning and reinforcement learning, and the architecture family. Overall, our results suggest a new use case for NCO solvers: representations learned for route construction can transfer to decision support.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑