Learning Invariants through Soft Unification
This addresses the challenge of automated invariant learning for AI systems, offering a novel method that is incremental in advancing neural network capabilities for reasoning tasks.
The paper tackles the problem of enabling machines to learn and use variables from examples without human engineering, proposing Unification Networks that lift examples into invariants through soft unification, and demonstrates improved performance over baselines on five datasets.
Human reasoning involves recognising common underlying principles across many examples. The by-products of such reasoning are invariants that capture patterns such as "if someone went somewhere then they are there", expressed using variables "someone" and "somewhere" instead of mentioning specific people or places. Humans learn what variables are and how to use them at a young age. This paper explores whether machines can also learn and use variables solely from examples without requiring human pre-engineering. We propose Unification Networks, an end-to-end differentiable neural network approach capable of lifting examples into invariants and using those invariants to solve a given task. The core characteristic of our architecture is soft unification between examples that enables the network to generalise parts of the input into variables, thereby learning invariants. We evaluate our approach on five datasets to demonstrate that learning invariants captures patterns in the data and can improve performance over baselines.