Learn to Explain Efficiently via Neural Logic Inductive Learning
arXiv:1910.02481
Abstract
The capability of making interpretable and self-explanatory decisions is essential for developing responsible machine learning systems. In this work, we study the learning to explain problem in the scope of inductive logic programming (ILP). We propose Neural Logic Inductive Learning (NLIL), an efficient differentiable ILP framework that learns first-order logic rules that can explain the patterns in the data. In experiments, compared with the state-of-the-art methods, we find NLIL can search for rules that are x10 times longer while remaining x3 times faster. We also show that NLIL can scale to large image datasets, i.e. Visual Genome, with 1M entities.
References in corpus (11)
- Categorical Reparameterization with Gumbel-Softmax
- Explaining Explanations in AI
- RotatE: Knowledge Graph Embedding by Relational Rotation in Complex Space
- Go for a Walk and Arrive at the Answer: Reasoning Over Paths in Knowledge Bases using Reinforcement Learning
- Traversing Knowledge Graphs in Vector Space
- Modeling Relation Paths for Representation Learning of Knowledge Bases
- Neural Logic Machines
- TensorLog: Deep Learning Meets Probabilistic DBs
- Inductive Logic Programming via Differentiable Deep Neural Logic Networks
- Logical Rule Induction and Theory Learning Using Neural Theorem Proving
- Towards Neural Theorem Proving at Scale