Provable Compositional Generalization for Object-Centric Learning
This provides a theoretical foundation for compositional generalization in object-centric learning, addressing a key gap in machine perception.
The paper tackles the problem of when object-centric representations guarantee compositional generalization, showing that autoencoders with specific structural assumptions and encoder-decoder consistency learn representations that provably generalize compositionally, validated on synthetic image data.
Learning representations that generalize to novel compositions of known concepts is crucial for bridging the gap between human and machine perception. One prominent effort is learning object-centric representations, which are widely conjectured to enable compositional generalization. Yet, it remains unclear when this conjecture will be true, as a principled theoretical or empirical understanding of compositional generalization is lacking. In this work, we investigate when compositional generalization is guaranteed for object-centric representations through the lens of identifiability theory. We show that autoencoders that satisfy structural assumptions on the decoder and enforce encoder-decoder consistency will learn object-centric representations that provably generalize compositionally. We validate our theoretical result and highlight the practical relevance of our assumptions through experiments on synthetic image data.