Provable Compositional Generalization for Object-Centric Learning

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.

Authors

• Thaddäus Wiedemer
• Jack Brady
• Alexander Panfilov
• Attila Juhos
• Matthias Bethge
• Wieland Brendel

Keywords

• compositional generalization
• identifiability
• object-centric learning
• generalization
• OOD generalization
• unsupervised learning
• slot attention
• disentanglement
• autoencoders
• representation learning