Learning by Analogy: A Causal Framework for Composition Generalization
By: Lingjing Kong , Shaoan Xie , Yang Jiao and more
Potential Business Impact:
Lets computers understand new ideas by breaking them down.
Compositional generalization -- the ability to understand and generate novel combinations of learned concepts -- enables models to extend their capabilities beyond limited experiences. While effective, the data structures and principles that enable this crucial capability remain poorly understood. We propose that compositional generalization fundamentally requires decomposing high-level concepts into basic, low-level concepts that can be recombined across similar contexts, similar to how humans draw analogies between concepts. For example, someone who has never seen a peacock eating rice can envision this scene by relating it to their previous observations of a chicken eating rice. In this work, we formalize these intuitive processes using principles of causal modularity and minimal changes. We introduce a hierarchical data-generating process that naturally encodes different levels of concepts and their interaction mechanisms. Theoretically, we demonstrate that this approach enables compositional generalization supporting complex relations between composed concepts, advancing beyond prior work that assumes simpler interactions like additive effects. Critically, we also prove that this latent hierarchical structure is provably recoverable (identifiable) from observable data like text-image pairs, a necessary step for learning such a generative process. To validate our theory, we apply insights from our theoretical framework and achieve significant improvements on benchmark datasets.
Similar Papers
Scalable Evaluation and Neural Models for Compositional Generalization
Machine Learning (CS)
Teaches computers to understand new things from old.
Scalable Evaluation and Neural Models for Compositional Generalization
Machine Learning (CS)
Teaches computers to understand new ideas.
Nonparametric Identification of Latent Concepts
Machine Learning (CS)
Teaches computers to learn like kids do.