Learning from interpretation transition using differentiable logic programming semantics

Learning from interpretation transition using differentiable logic programming semantics
复制标题

使用可微逻辑编程语义从解释转换中学习

DOI:
10.1007/s10994-021-06058-8
复制
发表时间:
2021-09
期刊:
影响因子:
7.5
通讯作者:
Katsumi Inoue
Katsumi Inoue
中科院分区:
计算机科学3区
文献类型:
--
作者:
Kun Gao;Hanpin Wang;Yongzhi Cao;Katsumi Inoue

文献摘要

参考文献

相似文献

学习与推理的结合是神经符号研究中一个重要而富有挑战性的课题。可微归纳逻辑规划是一种利用神经网络从完整、错误标记或不完整的观察事实中学习符号知识表示的技术。在本文中,我们提出了一种新的可微的归纳逻辑编程系统,称为可微学习的解释转换(D-LFIT),用于通过提出的逻辑程序、神经网络、优化算法的嵌入和一种计算逻辑程序语义的自适应代数方法来学习逻辑程序。提出的模型有几个特点,包括少量的参数,在课程学习设置中生成逻辑程序的能力,以及提取训练神经网络的线性时间复杂性。当系统从关系数据集学习时,引入了众所周知的底部子句定位算法。我们将我们的模型与NN-LFIT(从返回的连接网络中提取命题逻辑规则)、高度精确的规则学习器RIPPER、纯符号LFIT系统LF1T以及集成神经网络和命题化方法来处理一阶逻辑知识的cilp++进行了比较。从实验结果中,我们得出结论,当给定完整、不完整和错误标记的数据时,D-LFIT相对于基线产生相当的准确性。实验结果表明,D-LFIT不仅可以快速准确地学习符号逻辑程序,而且在处理错误标记和不完整数据集时也具有鲁棒性。
The combination of learning and reasoning is an essential and challenging topic in neuro-symbolic research. Differentiable inductive logic programming is a technique for learning a symbolic knowledge representation from either complete, mislabeled, or incomplete observed facts using neural networks. In this paper, we propose a novel differentiable inductive logic programming system called differentiable learning from interpretation transition (D-LFIT) for learning logic programs through the proposed embeddings of logic programs, neural networks, optimization algorithms, and an adapted algebraic method to compute the logic program semantics. The proposed model has several characteristics, including a small number of parameters, the ability to generate logic programs in a curriculum-learning setting, and linear time complexity for the extraction of trained neural networks. The well-known bottom clause positionalization algorithm is incorporated when the proposed system learns from relational datasets. We compare our model with NN-LFIT, which extracts propositional logic rules from retuned connected networks, the highly accurate rule learner RIPPER, the purely symbolic LFIT system LF1T, and CILP++, which integrates neural networks and the propositionalization method to handle first-order logic knowledge. From the experimental results, we conclude that D-LFIT yields comparable accuracy with respect to the baselines when given complete, incomplete, and mislabeled data. Our experimental results indicate that D-LFIT not only learns symbolic logic programs quickly and precisely but also performs robustly when processing mislabeled and incomplete datasets.
DOI: --
发表时间: 2019
期刊: FLAP
影响因子: --
作者:
Yin Jun Phua;Tony Ribeiro;Katsumi Inoue
通讯作者: Yin Jun Phua;Tony Ribeiro;Katsumi Inoue
DOI: --
发表时间: 1994
期刊: --
影响因子: --
作者:
Steffen Hölldobler;Y. Kalinke
通讯作者: Steffen Hölldobler;Y. Kalinke
DOI: 10.1145/321978.321991
发表时间: 1976-10
期刊: J. ACM
影响因子: --
作者:
M. H. Emden;R. Kowalski
通讯作者: M. H. Emden;R. Kowalski
DOI: 10.1016/j.jal.2004.03.003
发表时间: 2004-09
期刊: J. Appl. Log.
影响因子: --
作者:
Sebastian Bader;P. Hitzler
通讯作者: Sebastian Bader;P. Hitzler
DOI: --
发表时间: 2016-07
期刊: --
影响因子: --
作者:
William Yang Wang;William W. Cohen
通讯作者: William Yang Wang;William W. Cohen