Research map / Federated learning and privacy / Personalized and heterogeneous FL
Ditto: Fair and Robust Federated Learning Through Personalization
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 |
|---|---|---|---|
| Use Ditto personalization instead of the global model | federated learning · Across datasets and attacks | test accuracy on benign devices: ~6% (absolute) | Helped |
| Use Ditto personalization instead of TERM | fair federated learning · Clean data, across datasets | test accuracy variance across devices: ~10% | Helped |
| Use Ditto personalization instead of TERM | fair federated learning · Clean data, across datasets | test accuracy: 5% | Helped |
| Add gradient clipping to Ditto's global update aggregation (ablation) | Ditto · FEMNIST, A1 20% adversaries | test accuracy: .803 -> .810 | Helped |
| Add gradient clipping to Ditto's global update aggregation (ablation) | Ditto · FEMNIST, A1 80% adversaries | test accuracy: .669 -> .645 | Hurt |
| Add gradient clipping to Ditto's global update aggregation (ablation) | Ditto · FEMNIST, A2 20% adversaries | test accuracy: .792 -> .808 | Helped |
| Add gradient clipping to Ditto's global update aggregation (ablation) | Ditto · FEMNIST, A2 80% adversaries | test accuracy: .681 -> .684 | Helped |
| Add gradient clipping to Ditto's global update aggregation (ablation) | Ditto · FEMNIST, A3 10% adversaries | test accuracy: .695 -> .813 | Helped |
| Add gradient clipping to Ditto's global update aggregation (ablation) | Ditto · FEMNIST, A3 20% adversaries | test accuracy: .650 -> .672 | Helped |
| Replace joint optimization with local finetuning after obtaining the global model (ablation) | Ditto solver · Model replacement attack (A3), 20% adversaries | test accuracy | Hurt |
| Use approximate empirical Fisher or symmetrized KL instead of the L2 regularizer (ablation) | Ditto · Federated benchmark experiments | performance | No clear effect |
| Increase the TERM fairness parameter t (ablation) | TERM · Data poisoning (A1), increasing corruption | test accuracy on benign devices | Hurt |
Closest papers by meaning
- FairTrade: Achieving Pareto-Optimal Trade-Offs between Balanced Accuracy and Fairness in Federated Learning (AAAI 2024 · similarity 0.88)
- FedAA: A Reinforcement Learning Perspective on Adaptive Aggregation for Fair and Robust Federated Learning (AAAI 2025 · similarity 0.87)
- Fair Federated Learning under Domain Skew with Local Consistency and Domain Diversity (CVPR 2024 · similarity 0.86)
- Demystifying Local & Global Fairness Trade-offs in Federated Learning Using Partial Information Decomposition (ICLR 2024 · similarity 0.86)
- Towards the Robustness of Differentially Private Federated Learning (AAAI 2024 · similarity 0.86)
- FedFACT: A Provable Framework for Controllable Group-Fairness Calibration in Federated Learning (NeurIPS 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.