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 2412.14950cs.AI
收藏

在约束获取中泛化约束模型

Generalizing Constraint Models in Constraint Acquisition

KU Leuven(鲁汶大学);University College Cork(科克大学学院)

Dimos Tsouros, Senne Berden, Steven Prestwich, Tias Guns

AI总结:

针对约束获取无法泛化约束到参数化规范的问题,提出GenCon方法,利用统计学习训练分类器预测约束归属,提取决策规则或通过生成与测试方法构建可解释的参数化约束模型,实现高准确率与抗噪性。

摘要:

约束获取(CA)旨在通过协助用户进行建模过程来扩大约束编程的使用。然而,大多数CA方法存在一个显著的缺点:它们为特定问题实例学习一组单独的约束,但无法将这些约束泛化到该问题的参数化约束规范中。在本文中,我们通过提出GenCon来解决这一局限性,这是一种学习参数化约束模型的新方法,能够对同一问题的不同实例进行建模。为了实现这种泛化,我们在单个约束层面上利用了统计学习技术。具体而言,我们提议训练一个分类器,针对任何可能的约束和参数化,预测该约束是否属于该问题。然后我们展示了,对于某些类别的分类器,我们如何提取决策规则以构建可解释的约束规范。这使得能够为任何参数实例化生成基础约束。此外,我们提出了一种生成与测试方法,该方法可与任何分类器一起使用,以即时生成基础约束。我们的实验结果表明,我们的方法达到了高准确率,并且对输入实例中的噪声具有鲁棒性。

英文摘要:

Constraint Acquisition (CA) aims to widen the use of constraint programming by assisting users in the modeling process. However, most CA methods suffer from a significant drawback: they learn a single set of individual constraints for a specific problem instance, but cannot generalize these constraints to the parameterized constraint specifications of the problem. In this paper, we address this limitation by proposing GenCon, a novel approach to learn parameterized constraint models capable of modeling varying instances of the same problem. To achieve this generalization, we make use of statistical learning techniques at the level of individual constraints. Specifically, we propose to train a classifier to predict, for any possible constraint and parameterization, whether the constraint belongs to the problem. We then show how, for some classes of classifiers, we can extract decision rules to construct interpretable constraint specifications. This enables the generation of ground constraints for any parameter instantiation. Additionally, we present a generate-and-test approach that can be used with any classifier, to generate the ground constraints on the fly. Our empirical results demonstrate that our approach achieves high accuracy and is robust to noise in the input instances.

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。

↑