Research map / Probabilistic and scientific modeling / Molecules and proteins
Measuring Compositionality in Representation Learning
Jacob Andreas
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| Measure phrase compositionality with TRE rather than assuming it from word vectors | FastText CBOW phrase embeddings · Noun-noun compound human judgments | Spearman correlation with human compositionality ratings: ρ=−0.34, p<0.01 | Improved |
| Compare TRE across communication protocols | Speaker-listener RNNs · Reference game, held-out referents | correlation with generalization error: r=0.50, p<1e!−6 | Mixed |
| Compare TRE across communication protocols | Speaker-listener RNNs · Reference game, all training runs | correlation with absolute reward: r=0.57, p<1e!−9 | Hurt |
| Track TRE alongside mutual information during learning | Few-shot classifier · Color MNIST validation, training runs | correlation between TRE and mutual information: r=0.70, p < 1e!−10; r=0.71, p < 1e!−8 | Mixed |
| Restrict analysis to successful communication runs | Speaker-listener RNNs · Reference game, successful runs (held-out reward > 0.5) | correlation between TRE and generalization error: r=0.6, p<1e!−3 | Mixed |
| Restrict analysis to successful communication runs | Speaker-listener RNNs · Reference game, successful runs (held-out reward > 0.5) | correlation between TRE and absolute reward: r=0.38, p<0.05 | Hurt |
Closest papers by meaning
- Towards a Formal Theory of Representational Compositionality (ICML 2025 · similarity 0.89)
- When does compositional structure yield compositional generalization? A kernel theory. (ICLR 2025 · similarity 0.89)
- Does Data Scaling Lead to Visual Compositional Generalization? (ICML 2025 · similarity 0.86)
- Not Only Text: Exploring Compositionality of Visual Representations in Vision-Language Models (CVPR 2025 · similarity 0.86)
- Scalable Evaluation and Neural Models for Compositional Generalization (NeurIPS 2025 · similarity 0.86)
- How Do Language Models Compose Functions? (NeurIPS 2026 · similarity 0.85)
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.
