Neural Analogical Matching
This work addresses the challenge of creating data-efficient, human-like learning models in AI, though it appears incremental as part of a growing body of research on integrating analogy and deep learning.
The paper tackles the problem of integrating cognitive perspectives of analogy with deep learning by introducing the Analogical Matching Network, a neural architecture that learns to produce analogies between structured symbolic representations consistent with Structure-Mapping Theory, aiming for more robust and efficient learning techniques.
Analogy is core to human cognition. It allows us to solve problems based on prior experience, it governs the way we conceptualize new information, and it even influences our visual perception. The importance of analogy to humans has made it an active area of research in the broader field of artificial intelligence, resulting in data-efficient models that learn and reason in human-like ways. While cognitive perspectives of analogy and deep learning have generally been studied independently of one another, the integration of the two lines of research is a promising step towards more robust and efficient learning techniques. As part of a growing body of research on such an integration, we introduce the Analogical Matching Network: a neural architecture that learns to produce analogies between structured, symbolic representations that are largely consistent with the principles of Structure-Mapping Theory.