Research map / Detection and segmentation / Open-set and few-shot detection
Evaluating Disentanglement of Structured 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 |
|---|---|---|---|
| add disentanglement regularization (ablation) | GENESIS · Multi-dSprites | object-level disentanglement (%): 64 ± 3 -> 90 ± 0.8 | Helped |
| add disentanglement regularization (ablation) | IODINE · Multi-dSprites | object-level disentanglement (%): 35 ± 2 -> 70 ± 9 | Helped |
| add disentanglement regularization (ablation) | MONet (Att.) · Multi-dSprites | object-level disentanglement (%): 60 ± 4 -> 52 ± 4 | Hurt |
| add disentanglement regularization (ablation) | MONet (Dec.) · Multi-dSprites | object-level disentanglement (%): 69 ± 0.3 -> 75 ± 7 | Helped |
| add disentanglement regularization (ablation) | GENESIS · Multi-dSprites | ARI (%): 96 ± 0.6 -> 85 ± 0 | Hurt |
| add disentanglement regularization (ablation) | MONet (Att.) · Multi-dSprites | ARI (%): 98 ± 0.1 -> 95 ± 1 | Hurt |
| add disentanglement regularization (ablation) | MONet (Dec.) · Multi-dSprites | ARI (%): 91 ± 0.2 -> 80 ± 4 | Hurt |
| add disentanglement regularization (ablation) | GENESIS · CLEVR6 | object-level disentanglement (%): 43 ± 1 -> 50 ± 2 | Helped |
| add disentanglement regularization (ablation) | MONet (Att.) · CLEVR6 | object-level disentanglement (%): 40 ± 9 -> 46 ± 5 | Helped |
| add disentanglement regularization (ablation) | GENESIS · Multi-dSprites | property-level disentanglement (%): 65 ± 1 -> 72 ± 0.5 | Helped |
| use decoded masks (MONet Dec.) instead of attention masks (MONet Att.) | MONet · Multi-dSprites | object-level disentanglement (%): 52 ± 4 -> 75 ± 7 | Helped |
| use decoded masks (MONet Dec.) instead of attention masks (MONet Att.) | MONet · Multi-dSprites | ARI (%): 95 ± 1 -> 80 ± 4 | Hurt |
Closest papers by meaning
- Object centric architectures enable efficient causal representation learning (ICLR 2024 · similarity 0.88)
- Interaction Asymmetry: A General Principle for Learning Composable Abstractions (ICLR 2025 · similarity 0.88)
- Revisiting Disentanglement in Downstream Tasks: A Study on Its Necessity for Abstract Visual Reasoning (AAAI 2024 · similarity 0.88)
- Disentangled Representation Learning via Modular Compositional Bias (NeurIPS 2025 · similarity 0.88)
- Transferring disentangled representations: bridging the gap between synthetic and real images (NeurIPS 2024 · similarity 0.87)
- Disentangled 3D Scene Generation with Layout Learning (ICML 2024 · similarity 0.86)
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.