Samy Badreddine, Artur d'Avila Garcez, Luciano Serafini, Michael Spranger
TL;DR本文介绍了一种名为 Logic Tensor Networks(LTN)的神经符号形式和计算模型,支持通过引入一种称为 Real Logic 的可微分一阶逻辑表示语言进行学习和推理,并说明 LTN 提供了一种统一语言来规范和计算多个人工智能任务。
Abstract
Artificial Intelligence agents are required to learn from their surroundings
and to reason about the knowledge that has been learned in order to make
decisions. While state-of-the-art learning from data typically uses
sub-symbolic distributed representations, reasoning is normally useful at a
higher level of abstraction with the use of a first-order logic la