Research map / Label-efficient and robust learning / Recognition and long-tail
HyTrel: Hypergraph-enhanced Tabular Data Representation Learning
Pei Chen, Soumajyoti Sarkar, Leonard Lausen, Balasubramaniam Srinivasan, Sheng Zha, Ruihong Huang, George Karypis
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 |
|---|---|---|---|
| add ELECTRA pretraining (ablation) | HyTrel · TURL-CTA | F1 (%): 92.71 -> 93.53 | Improved |
| add ELECTRA pretraining (ablation) | HyTrel · TURL-CPA | F1 (%): 92.17 -> 93.48 | Improved |
| add ELECTRA pretraining (ablation) | HyTrel · table type detection | accuracy (%): 93.84 -> 95.81 | Improved |
| add ELECTRA pretraining (ablation) | HyTrel · PMC table similarity | F1 (%): 87.30 -> 87.32 | No clear effect |
| add contrastive pretraining (ablation) | HyTrel · PMC table similarity | F1 (%): 87.30 -> 89.26 | Improved |
| use contrastive objective instead of ELECTRA (ablation) | HyTrel · TURL-CTA | F1 (%): 93.53 -> 92.97 | Hurt |
| remove pretraining (ablation) | TaBERT (K=3) · TURL-CTA | F1 (%): 91.37 -> 88.97 | Hurt |
| remove pretraining (ablation) | TaBERT (K=1) · PMC table similarity | F1 (%): 86.18 -> 40.30 | Hurt |
| raise table size limit from (3, 2) to (30, 20) (ablation) | HyTrel ELECTRA · table type detection | test accuracy (%): 78.00 -> 95.81 | Improved |
| raise table size limit from (30, 20) to (240, 160) (ablation) | HyTrel ELECTRA · table type detection | test accuracy (%): 95.81 -> 95.67 | No clear effect |
| remove positional encodings (excessive invariance) (ablation) | TaBERT (K=1) · table type detection | test accuracy (%): 93.44 -> 88.91 | Hurt |
| remove positional encodings (excessive invariance) (ablation) | TaBERT (K=3) · table type detection | test accuracy (%): 95.22 -> 92.44 | Hurt |
Closest papers by meaning
- Large Scale Transfer Learning for Tabular Data via Language Modeling (NeurIPS 2024 · similarity 0.88)
- Enhancing Tabular Learners with Context-Aware Semantic Embeddings (NeurIPS 2026 · similarity 0.88)
- TabGLM: Tabular Graph Language Model for Learning Transferable Representations Through Multi-Modal Consistency Minimization (AAAI 2025 · similarity 0.88)
- ConTextTab: A Semantics-Aware Tabular In-Context Learner (NeurIPS 2025 · similarity 0.87)
- ICE-T: Interactions-aware Cross-column Contrastive Embedding for Heterogeneous Tabular Datasets (AAAI 2025 · similarity 0.87)
- UniTabE: A Universal Pretraining Protocol for Tabular Foundation Model in Data Science (ICLR 2024 · similarity 0.87)
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.
