Research map / Detection and segmentation / Open-set and few-shot detection
Provably Learning Object-Centric Representations
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| minimize both reconstruction error and compositional contrast in the inference model | inference model on synthetic data (independent and dependent latents) · synthetic multi-slot data (K in {2,3,5}) | Slot Identifiability Score | Helped |
| minimize compositional contrast (failing to do so lowers identifiability) (ablation) | inference model on synthetic data · synthetic multi-slot data | Slot Identifiability Score | Helped |
| lower reconstruction error and compositional contrast | existing object-centric models (MONet, Slot Attention, additive AE) · Spriteworld image data | Slot Identifiability Score | Helped |
| latent slot dimension larger than ground-truth (4 slots of dim 16 vs. ground-truth dim mismatch) | existing object-centric models on Spriteworld · Spriteworld image data | subtracted R^2 score in SIS computation | Hurt |
| dependent vs independent latents (ablation) | inference model on synthetic data · synthetic multi-slot data | Slot Identifiability Score | No clear effect |
Closest papers by meaning
- Provable Compositional Generalization for Object-Centric Learning (ICLR 2024 · similarity 0.92)
- Learning to Compose: Improving Object Centric Learning by Injecting Compositionality (ICLR 2024 · similarity 0.92)
- Compositional Scene Understanding through Inverse Generative Modeling (ICML 2025 · similarity 0.89)
- Unlocking Compositional Generalization in Continual Few-Shot Learning (NeurIPS 2026 · similarity 0.89)
- Object centric architectures enable efficient causal representation learning (ICLR 2024 · similarity 0.88)
- Temporally Consistent Object-Centric Learning by Contrasting Slots (CVPR 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.