Research map / Probabilistic and scientific modeling / Molecules and proteins
Probing the Compositionality of Intuitive Functions
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 |
|---|---|---|---|
| Replace radial basis or spectral mixture extrapolations with compositional kernel extrapolations | GP function extrapolation · Exp 1a forced choice, compositional ground truth | proportion chosen: 14% -> 69% | Helped |
| Change ground truth to spectral mixture generated functions | Compositional vs spectral extrapolation · Exp 1b forced choice | proportion chosen: 32% -> 68% | Helped |
| Use compositional marginal likelihood to predict MCMCP acceptance on generated functions | Compositional GP grammar · Exp 2a | correlation rho: 0.91 | Helped |
| Apply MCMCP to real-world data with periodic times RBF component | Compositional grammar · Exp 2b naturalistic functions | correlation rho: 0.83 | Helped |
| Compare spectral mixture to compositional kernel as predictor of drawings | Manual function completion · Exp 3 interpolation | RMSE: 86.96 -> 58.33 | Helped |
| Compare spectral mixture to compositional kernel as predictor of drawings | Manual function completion · Exp 3 extrapolation | RMSE: 110.45 -> 83.91 | Helped |
| Compare radial basis to compositional kernel as predictor of drawings | Manual function completion · Exp 3 interpolation | RMSE: 55.8 vs 58.33 | No clear effect |
| Compare radial basis to compositional kernel as predictor of drawings | Manual function completion · Exp 3 extrapolation | RMSE: 97.9 -> 83.91 | Helped |
| Use compositional rather than spectral mixture functions | Perceived predictability · Exp 4 isolated judgements | t statistic: t(948) = 11.422 | Helped |
| Use compositional rather than spectral mixture functions | Perceived predictability · Exp 4 paired comparisons | t statistic: t(499) = 13.502 | Helped |
| Increase number of observed points shown | Perceived predictability · Exp 4 | correlation with predictability: r = 0.23 | Helped |
| Increase number of observed points shown | Compositional vs spectral predictability gap · Exp 4 | correlation with gap: r = 0.14 | Helped |
Closest papers by meaning
- Towards a Formal Theory of Representational Compositionality (ICML 2025 · similarity 0.88)
- Nonparametric Identification of Latent Concepts (ICML 2025 · similarity 0.84)
- When does compositional structure yield compositional generalization? A kernel theory. (ICLR 2025 · similarity 0.83)
- CraftFactory: A Conditioned Control Policy Benchmark for Compositional Generalization (AAAI 2025 · similarity 0.83)
- Compositional Causal Reasoning Evaluation in Language Models (ICML 2025 · similarity 0.83)
- Diagnosing the Compositional Knowledge of Vision Language Models from a Game-Theoretic View (ICML 2024 · 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.