将当代人工智能技术应用于解决几何问题和自动演绎证明,一直是数学与人工智能这一交叉领域的重大挑战。这是我们系列研究工作的第四篇文章,在之前的工作中,我们建立了一个几何形式化系统,称为FormalGeo。此外,我们标注了大约7000个几何问题,构成了FormalGeo7k数据集。尽管FGPS(Formal Geometry Problem Solver)能够实现可解释的代数方程求解和类人演绎推理,但由于搜索策略的复杂性,它经常出现超时。在本文中,我们引入了FGeo-TP(Theorem Predictor),它利用语言模型预测用于解决几何问题的定理序列。我们比较了各种Transformer架构(如BART或T5)在定理预测中的有效性,在FGPS的搜索过程中实施剪枝,从而提高了其解决几何问题的性能。我们的结果表明,语言模型增强的FGeo-TP在FormalGeo7k数据集上的解题率显著提高,从39.7%上升到80.86%。此外,FGeo-TP在不同难度的问题上均表现出求解时间和搜索步骤的显著减少。
英文摘要:
The application of contemporary artificial intelligence techniques to address geometric problems and automated deductive proof has always been a grand challenge to the interdiscipline field of mathematics and artificial Intelligence. This is the fourth article in a series of our works, in our previous work, we established of a geometric formalized system known as FormalGeo. Moreover we annotated approximately 7000 geometric problems, forming the FormalGeo7k dataset. Despite the FGPS (Formal Geometry Problem Solver) can achieve interpretable algebraic equation solving and human-like deductive reasoning, it often experiences timeouts due to the complexity of the search strategy. In this paper, we introduced FGeo-TP (Theorem Predictor), which utilizes the language model to predict theorem sequences for solving geometry problems. We compared the effectiveness of various Transformer architectures, such as BART or T5, in theorem prediction, implementing pruning in the search process of FGPS, thereby improving its performance in solving geometry problems. Our results demonstrate a significant increase in the problem-solving rate of the language model-enhanced FGeo-TP on the FormalGeo7k dataset, rising from 39.7% to 80.86%. Furthermore, FGeo-TP exhibits notable reductions in solving time and search steps across problems of varying difficulty levels.
Variational Quantum Linear Solver enhanced Quantum Support Vector Machine
Fraunhofer Institute for Industrial Mathematics ITWM(弗劳恩霍夫工业数学研究所ITWM);RPTU Kaiserslautern-Landau(凯撒斯劳滕-兰道工业大学);Leibniz Universität Hannover(莱布尼茨汉诺威大学)
Solver-Informed RL: Grounding Large Language Models for Authentic Optimization Modeling
Cardinal Operations, China(中国卡迪纳尔运营公司);Shanghai University of Finance and Economics(上海财经大学);The University of Hong Kong(香港大学);Antai School of Economics and Management, Shanghai Jiao Tong University(上海交通大学安泰经济管理学院);Department of Management Science and Engineering, Stanford University(斯坦福大学管理科学与工程系)
Reinforcement Learning-based Non-Autoregressive Solver for Traveling Salesman Problems
College of Computer Science and Technology, Jilin University(吉林大学计算机科学与技术学院);Nanyang Technological University(南洋理工大学);School of Computer Science and Engineering, Nanyang Technological University(南洋理工大学计算机科学与工程学院);School of Computer Science, University of Science & Technology of China(中国科学技术大学计算机学院);School of Mathematical and Computer Sciences, Heriot-Watt University(赫瑞瓦特大学数学与计算机科学学院);College of Computer, National University of Defense Technology(国防科技大学计算机学院)
Left Heavy Tails and the Effectiveness of the Policy and Value Networks in DNN-based best-first search for Sokoban Planning
Cornell University(康奈尔大学)
尽管在SAT和CSP等各种NP完全领域的实际解决方案取得了成功,以及使用深度强化学习来解决围棋等双人游戏,但某些类别的PSPACE硬规划问题仍然遥不可及。由于硬实例上的指数搜索空间,即使是精心设计的领域专用求解器也可能很快失败。最近的工作结合传统的搜索方法,如最佳优先搜索和蒙特卡罗树搜索,与深度神经网络(DNN)算法已经显示出有前途的进展,可以解决大量的硬规划实例超出了专门的求解器。为了更好地理解为什么这些方法有效,我们研究了基于DNN的最佳优先搜索在Sokoban上的策略和价值网络的相互作用,并展示了策略网络的惊人有效性,价值网络进一步增强了搜索的指导启发。为了进一步理解这种现象,我们研究了搜索算法的成本分布,发现Sokoban实例可以具有重尾运行时分布,尾部位于左侧和右侧。特别是,第一次,我们表明存在的\textit{left heavy tails},并提出了一个抽象的树模型,可以凭经验解释这些尾巴的外观。实验表明,作为一个强大的启发式引导搜索,这可以导致左重尾多项式缩放,避免探索指数大小的子树的策略网络的关键作用。我们的研究结果还证明了随机重新启动的重要性,正如在传统的组合求解器中广泛使用的那样,对于基于DNN的搜索方法,以避免左和右重尾。