Research map / Label-efficient and robust learning / Recognition and long-tail
The Semantic Architect: How FEAML Bridges Structured Data and LLMs for Multi-Label Tasks
Wanfu Gao, Zebin He, Jun Gao
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 |
|---|---|---|---|
| use FEAML features instead of no feature generation (Base) | downstream classifier · adult | Accuracy: 0.6183 -> 0.6312 | Improved |
| use FEAML instead of CAAFE | FEAML vs CAAFE · communities | Accuracy: 0.5301 -> 0.5478 | Improved |
| use FEAML instead of CAAFE | FEAML vs CAAFE · myocardial | Accuracy: 0.1395 -> 0.1355 | Hurt |
| use FEAML instead of FeatLLM | FEAML vs FeatLLM · myocardial | Hamming Loss: 0.1275 -> 0.1277 | No clear effect |
| use FEAML instead of FeatLLM | FEAML vs FeatLLM · bank | F1-score (micro): 0.6534 -> 0.6724 | Improved |
| use FEAML instead of Base with Random Forest | FEAML · adult (RF downstream) | Accuracy: 0.6062 -> 0.6602 | Improved |
| use FEAML instead of Base with MLKNN | FEAML · adult (MLKNN downstream) | Accuracy: 0.5469 -> 0.6693 | Improved |
| use FEAML instead of AFAT with RF | FEAML vs AFAT · adult (RF downstream) | Accuracy: 0.5945 -> 0.6602 | Improved |
| add structured metadata prompts (ablation) | FEAML · adult, bank, communities, credit-g | Hamming Loss | Improved |
| add label co-occurrence prompts (ablation) | FEAML · adult, bank, communities, credit-g | Hamming Loss | Improved |
Closest papers by meaning
- Learning Compact Model for Large-Scale Multi-Label Data (AAAI 2019 · similarity 0.85)
- Distilling Reliable Knowledge for Instance-Dependent Partial Label Learning (AAAI 2024 · similarity 0.84)
- Handling Learnwares from Heterogeneous Feature Spaces with Explicit Label Exploitation (NeurIPS 2024 · similarity 0.83)
- In-Context Learning Learns Label Relationships but Is Not Conventional Learning (ICLR 2024 · similarity 0.83)
- Heterogeneous Label Shift: Theory and Algorithm (ICML 2025 · similarity 0.83)
- Specifying What You Know or Not for Multi-Label Class-Incremental Learning (AAAI 2025 · 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.
