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 2009.05908cs.LGstat.ML
收藏

理解深度神经网络上的布尔函数可学习性:PAC学习与神经符号模型的结合

Understanding Boolean Function Learnability on Deep Neural Networks: PAC Learning Meets Neurosymbolic Models

Institute of Informatics, Federal University of Rio Grande do Sul(里约格兰德杜斯阿尔加维联邦大学信息学院);Federal University of Rio Grande do Sul(里约格兰德杜斯阿尔加维联邦大学);Department of Computer Science, Rice University(里士大学计算机科学系)

Marcio Nicolau, Anderson R. Tavares, Zhiwei Zhang, Pedro Avelar, João M. Flach, Luis C. Lamb, Moshe Y. Vardi

AI总结:

本文研究深度神经网络如何学习布尔公式,发现神经学习比纯符号方法泛化更好,且小而浅的网络能有效近似组合优化相关公式,为可解释神经符号AI提供了新见解。

摘要:

计算学习理论指出,许多布尔公式类可在多项式时间内学习。本文研究了实践中此类公式如何由深度神经网络学习这一被忽视的问题。具体而言,我们分析了与模型采样基准、组合优化问题以及具有不同约束度的随机3-CNF相关的布尔公式。实验表明:(i) 神经学习比纯规则系统和纯符号方法泛化能力更好;(ii) 相对较小且浅层的神经网络是组合优化问题相关公式的良好近似器;(iii) 较小的公式似乎更难学习,可能由于可用的正(满足)样本较少;(iv) 有趣的是,欠约束的3-CNF公式比过约束的公式更具学习挑战性。这些发现为更好地理解、构建和使用可解释的神经符号AI方法铺平了道路。

英文摘要:

Computational learning theory states that many classes of boolean formulas are learnable in polynomial time. This paper addresses the understudied subject of how, in practice, such formulas can be learned by deep neural networks. Specifically, we analyze boolean formulas associated with model-sampling benchmarks, combinatorial optimization problems, and random 3-CNFs with varying degrees of constrainedness. Our experiments indicate that: (i) neural learning generalizes better than pure rule-based systems and pure symbolic approach; (ii) relatively small and shallow neural networks are very good approximators of formulas associated with combinatorial optimization problems; (iii) smaller formulas seem harder to learn, possibly due to the fewer positive (satisfying) examples available; and (iv) interestingly, underconstrained 3-CNF formulas are more challenging to learn than overconstrained ones. Such findings pave the way for a better understanding, construction, and use of interpretable neurosymbolic AI methods.

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。

↑