Research map / Vision-language and multimodal / Human and embodied modeling
Stratified Rule-Aware Network for Abstract Visual Reasoning
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 SRAN's stratified rule embeddings and gated fusion | RPM solving · I-RAVEN | accuracy: 46.1 -> 60.8 | Helped |
| Use SRAN's stratified rule embeddings and gated fusion | RPM solving · PGM | accuracy: 68.1 -> 71.3 | Helped |
| Generate balanced answer sets with ABT rather than the original RAVEN answer sets | RPM benchmark evaluation · RAVEN and I-RAVEN | accuracy: 91.4 -> 46.1; 94.2 -> 14.2 | Mixed |
| Include column-rule induction during training (ablation) | SRAN · I-RAVEN | accuracy: 60.8 -> 59.6 | Hurt |
| Preserve order in cell-wise embeddings instead of summing them orderlessly (ablation) | SRAN · I-RAVEN | accuracy: 23.5 -> 36.7 | Helped |
| Add the individual-wise hierarchy to the cell-wise hierarchy (ablation) | SRAN · I-RAVEN | accuracy: 36.7 -> 57.8 | Helped |
| Add the ecological hierarchy to cell-wise and individual-wise hierarchies (ablation) | SRAN · I-RAVEN | accuracy: 57.8 -> 60.8 | Helped |
| Use all three hierarchies rather than cell-wise and ecological hierarchies only (ablation) | SRAN · I-RAVEN | accuracy: 52.9 -> 60.8 | Helped |
| Use all three hierarchies rather than individual-wise and ecological hierarchies only (ablation) | SRAN · I-RAVEN | accuracy: 57.0 -> 60.8 | Helped |
| Use all three hierarchies rather than any single hierarchy (ablation) | SRAN · I-RAVEN | accuracy: 36.7 -> 60.8; 48.7 -> 60.8; 51.6 -> 60.8 | Helped |
Closest papers by meaning
- Towards Generative Abstract Reasoning: Completing Raven’s Progressive Matrix via Rule Abstraction and Selection (ICLR 2024 · similarity 0.92)
- Learning Visual Abstract Reasoning through Dual-Stream Networks (AAAI 2024 · similarity 0.88)
- SLVR: Structured Latent Visual Reasoning via Human-like Reasoning Flows (NeurIPS 2026 · similarity 0.84)
- StemBind: When MLLMs Get Lost Between Rules and Instances in Abstract Visual Reasoning (NeurIPS 2026 · similarity 0.84)
- GenVP: Generating Visual Puzzles with Contrastive Hierarchical VAEs (ICLR 2025 · similarity 0.84)
- Boosting Inference with Guided Reasoning: Stochastic Exploration for Recursive Models (NeurIPS 2026 · similarity 0.83)
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.