Towards bridging the neuro-symbolic gap: deep deductive reasoners
Towards bridging the neuro-symbolic gap: deep deductive reasoners
复制标题
DOI：
10.1007/s10489-020-02165-6
复制
发表时间：
2021-02
影响因子：
5.3
通讯作者：
Monireh Ebrahimi;Aaron Eberhart;Federico Bianchi;P. Hitzler
中科院分区：
文献类型：
--
作者：
Monireh Ebrahimi;Aaron Eberhart;Federico Bianchi;P. Hitzler
符号知识表示和推理以及深度学习是人工智能的根本不同方法，具有互补的功能。前者是透明的和数据有效的，但它们对噪声敏感，不能应用于数据不明确的非符号域。后者可以从示例中学习复杂的任务，对噪声具有鲁棒性，但它是黑箱;需要大量数据（不一定容易获得），学习速度慢，并且容易出现对抗性示例。任何一种范式都擅长于某些类型的问题，而另一种范式表现不佳。为了开发更强大的AI系统，正在寻求将联合收割机人工神经网络和符号推理相结合的集成神经符号系统。在这种情况下，一个基本的开放问题是如何执行基于逻辑的演绎推理的知识库，通过可训练的人工神经网络。本文简要总结了作者最近在深度演绎推理的背景下为弥合神经和符号鸿沟所做的努力。在本文中，我们将讨论模型在准确性，可扩展性，可转移性，可推广性，速度和可解释性方面的优势和局限性，最后，将讨论可能的修改，以增强所需的功能。更具体地说，在架构方面，我们正在研究内存增强网络、逻辑张量网络和LSTM模型的组合，以探索它们在进行演绎推理方面的能力和局限性。我们将这些模型分别应用于资源描述框架（RDF）、一阶逻辑和描述逻辑。
Symbolic knowledge representation and reasoning and deep learning are fundamentally different approaches to artificial intelligence with complementary capabilities. The former are transparent and data-efficient, but they are sensitive to noise and cannot be applied to non-symbolic domains where the data is ambiguous. The latter can learn complex tasks from examples, are robust to noise, but are black boxes; require large amounts of –not necessarily easily obtained– data, and are slow to learn and prone to adversarial examples. Either paradigm excels at certain types of problems where the other paradigm performs poorly. In order to develop stronger AI systems, integrated neuro-symbolic systems that combine artificial neural networks and symbolic reasoning are being sought. In this context, one of the fundamental open problems is how to perform logic-based deductive reasoning over knowledge bases by means of trainable artificial neural networks. This paper provides a brief summary of the authors’ recent efforts to bridge the neural and symbolic divide in the context of deep deductive reasoners. Throughout the paper we will discuss strengths and limitations of models in term of accuracy, scalability, transferability, generalizabiliy, speed, and interpretability, and finally, will talk about possible modifications to enhance desirable capabilities. More specifically, in terms of architectures, we are looking at Memory-augmented networks, Logic Tensor Networks, and compositions of LSTM models to explore their capabilities and limitations in conducting deductive reasoning. We are applying these models on Resource Description Framework (RDF), first-order logic, and the description logicrespectively.