Research map / Detection and segmentation / Open-set and few-shot detection
Relational Neural Expectation Maximization: Unsupervised Discovery of Objects and their Interactions
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 |
|---|---|---|---|
| Use pairwise relational interactions between object representations | R-NEM · Bouncing balls with invisible curtain | BCE: 94.64 -> 46.22 | Helped |
| Use pairwise relational interactions between object representations | R-NEM · Bouncing balls with invisible curtain | relational BCE: 4.14 -> 2.33 | Helped |
| Use pairwise relational interactions between object representations | R-NEM · Bouncing balls with invisible curtain | BCE: 59.32 -> 46.22 | Helped |
| Use pairwise relational interactions between object representations | R-NEM · Bouncing balls with invisible curtain | relational BCE: 2.72 -> 2.33 | Helped |
| Represent object assignments with compositional components | R-NEM · Bouncing balls | ARI: 0.8 | Helped |
| Add attention to select relevant context objects (ablation) | R-NEM · Bouncing balls, 4-ball test set | BCE | No clear effect |
| Add attention to select relevant context objects (ablation) | R-NEM · Bouncing balls, 6-8-ball test set | BCE | Helped |
| Increase the number of components during training (ablation) | R-NEM · Bouncing balls | grouping quality | Helped |
| Increase the number of components during training (ablation) | R-NEM · Bouncing balls | relational BCE | Helped |
Closest papers by meaning
- Hierarchical Relational Inference (AAAI 2021 · similarity 0.88)
- Reasoning About Physical Interactions with Object-Oriented Prediction and Planning (ICLR 2019 · similarity 0.88)
- Physical Object Understanding with a Physically Controllable World Model (CVPR 2026 · similarity 0.87)
- Unsupervised Discovery of 3D Physical Objects from Video (ICLR 2021 · similarity 0.86)
- Neural Expectation Maximization (NeurIPS 2017 · similarity 0.86)
- Object centric architectures enable efficient causal representation learning (ICLR 2024 · similarity 0.84)
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.