arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2603.20801cs.LG
收藏

大邻域搜索与迭代神经约束启发式方法的结合

Large Neighborhood Search meets Iterative Neural Constraint Heuristics

Department of Mechanical \& Industrial Engineering, University of Toronto Vector Institute

Yudong W. Xu, Wenhao Li, Scott Sanner, Elias B. Khalil

AI总结:

本文将迭代神经启发式与大邻域搜索结合,改进ConsFormer方法,通过实验发现该方法在Sudoku、图着色和MaxCut任务中性能提升显著,揭示了LNS在设计迭代神经方法中的有效性。

摘要:

神经网络越来越多地被用作约束满足问题的启发式方法。这些神经方法通常具有递归性,学习迭代地优化候选分配。本文明确迭代神经启发式与大邻域搜索(LNS)之间的联系,并将现有的神经约束满足方法ConsFormer适应到LNS过程中。我们将得到的神经LNS分解为两个标准组件:破坏和修复操作符。在破坏侧,我们实例化了几种经典启发式方法,并引入了新的基于预测的运算符,利用模型内部分数选择邻域。在修复侧,我们利用ConsFormer作为神经修复运算符,并将原始基于采样的解码器与选择最可能分配的贪心解码器进行比较。通过在数独、图着色和MaxCut上的实证研究,我们发现将神经启发式适应到LNS过程中在原始设置上取得了显著的提升,并提高了其与经典和神经基线的竞争力。我们进一步观察到在任务中存在一致的设计模式:随机破坏运算符优于贪心运算符,而贪心修复比基于采样的修复更有效于找到单一高质量可行分配。这些发现突显了LNS作为结构和改进迭代神经方法的有用透镜和设计框架。

英文摘要:

Neural networks are being increasingly used as heuristics for constraint satisfaction. These neural methods are often recurrent, learning to iteratively refine candidate assignments. In this work, we make explicit the connection between such iterative neural heuristics and Large Neighborhood Search (LNS), and adapt an existing neural constraint satisfaction method-ConsFormer-into an LNS procedure. We decompose the resulting neural LNS into two standard components: the destroy and repair operators. On the destroy side, we instantiate several classical heuristics and introduce novel prediction-guided operators that exploit the model's internal scores to select neighborhoods. On the repair side, we utilize ConsFormer as a neural repair operator and compare the original sampling-based decoder to a greedy decoder that selects the most likely assignments. Through an empirical study on Sudoku, Graph Coloring, and MaxCut, we find that adapting the neural heuristic to an LNS procedure yields substantial gains over its vanilla settings and improves its competitiveness with classical and neural baselines. We further observe consistent design patterns across tasks: stochastic destroy operators outperform greedy ones, while greedy repair is more effective than sampling-based repair for finding a single high-quality feasible assignment. These findings highlight LNS as a useful lens and design framework for structuring and improving iterative neural approaches.

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。

↑