Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Conceptualized and Contextualized Gaussian Embedding
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 |
|---|---|---|---|
| Represent words as sememe-aggregated Gaussian distributions with symmetric and asymmetric training objectives | TIGER · Word similarity average | Spearman correlation (ρ): 64.50 -> 70.20 | Helped |
| Represent words as sememe-aggregated Gaussian distributions with symmetric and asymmetric training objectives | TIGER · Word entailment average | F1: 77.56 -> 80.40 | Helped |
| Add GIANT Gaussian contextualization (ablation) | TIGER · POS | F1-score (%): 67.20 -> 81.46 | Helped |
| Add HowNet sememe information (ablation) | TIGER + GIANT · POS | F1-score (%): 76.88 -> 81.46 | Helped |
| Add Gaussian Distribution Attention (ablation) | TIGER + GIANT · POS | F1-score (%): 79.66 -> 81.46 | Helped |
| Sample sememe-sharing words in addition to window-based contexts (ablation) | TIGER + GIANT · POS | F1-score (%): 79.89 -> 81.46 | Helped |
| Retain uncertainty information at inference (ablation) | TIGER + GIANT · POS | F1-score (%): 81.40 -> 81.46 | No clear effect |
| Retain uncertainty information at inference (ablation) | TIGER + GIANT · STC (WeBis) | F1-score (%): 77.50 -> 77.50 | No clear effect |
| Concatenate conceptualized and contextualized Gaussian representations | BERT · Word similarity average | Spearman correlation (ρ): 32.22 -> 63.36 | Helped |
| Concatenate conceptualized and contextualized Gaussian representations | BERT · POS | F1-score (%): 87.67 -> 90.20 | Helped |
Closest papers by meaning
- Geometric Relationship between Word and Context Representations (AAAI 2018 · similarity 0.88)
- Context-guided Embedding Adaptation for Effective Topic Modeling in Low-Resource Regimes (NeurIPS 2023 · similarity 0.88)
- Learning Semantic Representations for Novel Words: Leveraging Both Form and Context (AAAI 2019 · similarity 0.88)
- Learning Context-Specific Word/Character Embeddings (AAAI 2017 · similarity 0.88)
- Beyond Single Concept Vector: Modeling Concept Subspace in LLMs with Gaussian Distribution (ICLR 2025 · similarity 0.87)
- Contextual Document Embeddings (ICLR 2025 · 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.