Research map / Federated learning and privacy / Personalized and heterogeneous FL
Towards Understanding and Mitigating Dimensional Collapse in Heterogeneous Federated Learning
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 |
|---|---|---|---|
| Add correlation-matrix Frobenius-norm regularization during local training | FedAvg on CIFAR10 · CIFAR10 α=0.05 | test accuracy (%): 64.85 -> 73.06 | Helped |
| Add correlation-matrix Frobenius-norm regularization during local training | FedAvg on CIFAR100 · CIFAR100 α=0.05 | test accuracy (%): 59.87 -> 61.53 | Helped |
| Add correlation-matrix Frobenius-norm regularization during local training | FedAvg on TinyImageNet · TinyImageNet α=0.05 | test accuracy (%): 35.02 -> 40.29 | Helped |
| Add FedDecorr in the homogeneous setting | FedAvg on CIFAR10 · CIFAR10 α=∞ | test accuracy (%): 92.39 -> 92.19 | Hurt |
| Add FedDecorr in the homogeneous setting | FedAvg on TinyImageNet · TinyImageNet α=∞ | test accuracy (%): 49.33 -> 52.63 | Helped |
| Regularize the covariance matrix rather than the correlation matrix (ablation) | FedAvg on TinyImageNet · TinyImageNet α=0.05 | test accuracy (%): 35.02 -> 32.88 | Hurt |
| Use Structured-DeCov instead of correlation-matrix regularization (ablation) | FedAvg on TinyImageNet · TinyImageNet α=0.05 | test accuracy (%): 35.02 -> 32.04 | Hurt |
| Increase number of clients to 100 and add FedDecorr (ablation) | FedAvg on TinyImageNet · TinyImageNet α=0.05 | test accuracy (%): 21.53 -> 30.55 | Helped |
| Vary local epochs from 1 to 20 while using FedDecorr (ablation) | FedAvg on CIFAR100 · CIFAR100 α=0.05 | test accuracy (%): 53.18; 60.65 | Mixed |
Closest papers by meaning
- Widening the Network Mitigates the Impact of Data Heterogeneity on FedAvg (ICML 2025 · similarity 0.88)
- Rethinking the Representation in Federated Unsupervised Learning with Non-IID Data (CVPR 2024 · similarity 0.88)
- FedSMU: Communication-Efficient and Generalization-Enhanced Federated Learning through Symbolic Model Updates (ICML 2025 · similarity 0.88)
- FedGPS: Statistical Rectification Against Data Heterogeneity in Federated Learning (NeurIPS 2025 · similarity 0.88)
- FedLMT: Tackling System Heterogeneity of Federated Learning via Low-Rank Model Training with Theoretical Guarantees (ICML 2024 · similarity 0.88)
- FedLoGe: Joint Local and Generic Federated Learning under Long-tailed Data (ICLR 2024 · similarity 0.88)
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.