Research map / Probabilistic and scientific modeling / Molecules and proteins
Raven’s Progressive Matrices Completion with Latent Gaussian Process Priors
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 latent Gaussian process priors for concept-changing rules instead of VAE-GAN inpainting | RPM-like panel completion · Polygon | MSE: 5.16e-2 -> 7.55e-3 ± 3.9e-5 | Helped |
| Use latent Gaussian process priors for concept-changing rules instead of VAE-GAN inpainting | RPM-like panel completion · Circle | MSE: 1.08e-1 -> 6.43e-3 ± 1.8e-5 | Helped |
| Use latent Gaussian process priors for concept-changing rules instead of VAE-GAN inpainting | RPM-like panel completion · Square | MSE: 4.31e-2 -> 8.00e-3 ± 2.6e-5 | Helped |
| Use latent Gaussian process priors for concept-changing rules instead of VAE-GAN inpainting | RPM-like panel completion · Complex Polygon | MSE: 3.47e-2 -> 8.27e-3 ± 3.1e-5 | Helped |
| Use latent Gaussian process priors for concept-changing rules instead of VAE-GAN inpainting | RPM-like panel completion · Complex Circle | MSE: 1.98e-1 -> 1.47e-2 ± 5.2e-5 | Helped |
| Use latent Gaussian process priors for concept-changing rules instead of VAE-GAN inpainting | RPM-like panel completion · Position Polygon (triangle) | MSE: 2.40e-2 -> 9.22e-3 ± 1.7e-5 | Helped |
| Use latent Gaussian process priors for concept-changing rules instead of VAE-GAN inpainting | RPM-like panel completion · Position Polygon (square) | MSE: 3.55e-2 -> 9.28e-3 ± 2.1e-5 | Helped |
| Use latent Gaussian process priors for concept-changing rules instead of VAE-GAN inpainting | RPM-like panel completion · Position Circle | MSE: 1.68e-2 -> 1.12e-2 ± 1.7e-5 | Helped |
| Use latent GP priors rather than a β-VAE objective alone | Concept-specific representations · Circle | Factor VAE Score: 7.59e-1 ± 1.8e-2 -> 1.00 ± 0.0 | Helped |
| Use latent GP priors rather than a β-VAE objective alone | Concept-specific representations · Polygon | SAP Score: 4.40e-1 ± 6.5e-3 -> 7.80e-1 ± 3.8e-3 | Helped |
| Use latent GP priors rather than a β-VAE objective alone | Concept-specific representations · Circle | SAP Score: 9.41e-2 ± 3.0e-3 -> 7.93e-1 ± 4.0e-3 | Helped |
| Train with fewer examples | Proposed model · Polygon | MSE | Mixed |
Closest papers by meaning
- Towards Generative Abstract Reasoning: Completing Raven’s Progressive Matrix via Rule Abstraction and Selection (ICLR 2024 · similarity 0.91)
- GenVP: Generating Visual Puzzles with Contrastive Hierarchical VAEs (ICLR 2025 · similarity 0.83)
- Latent Thinking Optimization: Your Latent Reasoning Language Model Secretly Encodes Reward Signals in its Latent Thoughts (ICLR 2026 · similarity 0.83)
- MARVEL: Multidimensional Abstraction and Reasoning through Visual Evaluation and Learning (NeurIPS 2024 · similarity 0.83)
- Tracing the Traces: Latent Temporal Signals for Efficient and Accurate Reasoning (ICLR 2026 · similarity 0.83)
- Semantic-Enriched Latent Visual Reasoning (ICML 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.