Research map / Federated learning and privacy / Personalized and heterogeneous FL
Neuro-Symbolic Federated Learning over Heterogeneous Data-Views: A Structured Approach to Distributive EHR Modelling
Soheila Molaei, Bahareh Fatemi, Anshul Thakur, Andrew Soltan, Fazle Rabbi, Andreas L. Opdahl, Kim Branson, Patrick Schwab, Danielle Belgrave, David A. Clifton
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 |
|---|---|---|---|
| replace flat aligned FedAvg with knowledge-graph NS-FL | NS-FL · MC-MED (average over clients) | AUROC: 0.700 ± 0.011 -> 0.795 ± 0.004 | Improved |
| use NS-FL instead of Knowledge Filtering | NS-FL · CURIAL (data-view heterogeneity) | AUROC: +1.5% relative | Improved |
| use NS-FL instead of Knowledge Filtering | NS-FL · eICU (data-view heterogeneity) | AUPRC: +22.4% relative | Improved |
| use NS-FL instead of Knowledge Filtering | NS-FL · MIMIC-III (data-view heterogeneity) | AUPRC: +9.5% relative | Improved |
| use NS-FL instead of Knowledge Filtering on a client with fewer modalities | NS-FL · MC-MED Client 3 | AUROC: 0.765 ± 0.015 -> 0.851 ± 0.003 | Improved |
| use NS-FL instead of Knowledge Filtering on a client with all modalities | NS-FL · MC-MED Client 1 | AUROC: 0.804 ± 0.009 -> 0.813 ± 0.004 | Improved |
| use NS-FL in standard aligned feature spaces | NS-FL · CURIAL, eICU, MIMIC-III (standard) | AUROC/AUPRC | No clear effect |
| align feature views manually for FedAvg under heterogeneity (ablation) | FedAvg · CURIAL, eICU, MIMIC-III (heterogeneous) | AUROC/AUPRC | Hurt |
| disable learned embeddings and relation attention (non-learnable symbolic graph) (ablation) | NS-FL · CURIAL | AUROC | Hurt |
| keep semantic typing instead of homogeneous graph (ablation) | NS-FL · CURIAL | AUROC | Improved |
Closest papers by meaning
- MH-pFLID: Model Heterogeneous personalized Federated Learning via Injection and Distillation for Medical Data Analysis (ICML 2024 · similarity 0.87)
- FedVCK: Non-IID Robust and Communication-Efficient Federated Learning via Valuable Condensed Knowledge for Medical Image Analysis (AAAI 2025 · similarity 0.86)
- dFLMoE: Decentralized Federated Learning via Mixture of Experts for Medical Data Analysis (CVPR 2025 · similarity 0.86)
- Incongruent Multimodal Federated Learning for Medical Vision and Language-based Multi-label Disease Detection (AAAI 2025 · similarity 0.85)
- FEDMEKI: A Benchmark for Scaling Medical Foundation Models via Federated Knowledge Injection (NeurIPS 2024 · similarity 0.84)
- Federated Learning with Heterogeneous LLMs: Integrating Small Student Client Models with a Large Hungry Model (AAAI 2025 · similarity 0.84)
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.
