Identifiable Object-Centric Representation Learning via Probabilistic Slot Attention
This work addresses the lack of theoretical identifiability guarantees in object-centric representation learning, which is crucial for scaling slot-based methods to high-dimensional images with correctness guarantees, representing an incremental advancement in the field.
The paper tackles the problem of learning modular object-centric representations for systematic generalization by proposing a probabilistic slot-attention algorithm that imposes an aggregate mixture prior, providing theoretical identifiability guarantees up to an equivalence relation without supervision. The result is empirically verified on 2D data and high-resolution imaging datasets.
Learning modular object-centric representations is crucial for systematic generalization. Existing methods show promising object-binding capabilities empirically, but theoretical identifiability guarantees remain relatively underdeveloped. Understanding when object-centric representations can theoretically be identified is crucial for scaling slot-based methods to high-dimensional images with correctness guarantees. To that end, we propose a probabilistic slot-attention algorithm that imposes an aggregate mixture prior over object-centric slot representations, thereby providing slot identifiability guarantees without supervision, up to an equivalence relation. We provide empirical verification of our theoretical identifiability result using both simple 2-dimensional data and high-resolution imaging datasets.