Learning Structured Declarative Rule Sets -- A Challenge for Deep Discrete Learning
This paper highlights a foundational problem in inductive rule learning, specifically the inability to autonomously form structured rule bases, which impedes progress for researchers in this domain.
This paper identifies the challenge of learning structured declarative rule sets, where inputs are combined to form new auxiliary concepts for subsequent rules. The authors argue that the lack of this capability in current inductive rule learning is a major obstacle to progress in the field.
Arguably the key reason for the success of deep neural networks is their ability to autonomously form non-linear combinations of the input features, which can be used in subsequent layers of the network. The analogon to this capability in inductive rule learning is to learn a structured rule base, where the inputs are combined to learn new auxiliary concepts, which can then be used as inputs by subsequent rules. Yet, research on rule learning algorithms that have such capabilities is still in their infancy, which is - we would argue - one of the key impediments to substantial progress in this field. In this position paper, we want to draw attention to this unsolved problem, with a particular focus on previous work in predicate invention and multi-label rule learning