Research map / Federated learning and privacy / Personalized and heterogeneous FL
Understanding and Mitigating Dimensional Collapse in Federated Learning
Yujun Shi, Jian Liang, Wenqing Zhang, Chuhui Xue, Vincent Y. F. Tan, Song Bai
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 FedDecorr regularizer | FedAvg · CIFAR10 (alpha=0.05) | test accuracy (%): 64.85 -> 73.06 | Improved |
| add FedDecorr regularizer | FedAvg · CIFAR100 (alpha=0.05) | test accuracy (%): 59.87 -> 61.53 | Improved |
| add FedDecorr regularizer | FedAvg · TinyImageNet (alpha=0.05) | test accuracy (%): 35.02 -> 40.29 | Improved |
| add FedDecorr regularizer | FedAvg · CIFAR10 (alpha=0.5) | test accuracy (%): 89.84 -> 89.84 | No clear effect |
| add FedDecorr regularizer | FedAvg · CIFAR100 (homogeneous, alpha=inf) | test accuracy (%): 74.54 -> 73.87 | No clear effect |
| add FedDecorr regularizer | FedAvg · TinyImageNet (homogeneous, alpha=inf) | test accuracy (%): 49.33 -> 52.63 | Improved |
| increase number of clients to 100 with FedDecorr | FedAvg · TinyImageNet (100 clients, alpha=0.05) | test accuracy (%): 21.53 -> 30.55 | Improved |
| add FedDecorr on wider vs narrower networks | FedAvg · CIFAR10 (alpha=0.05) | test accuracy (%): ResNet18 71.51 -> 76.54; ResNet32 65.76 -> 67.21 | Mixed |
| use DeCov covariance regularization instead of FedDecorr correlation regularization (ablation) | FedAvg · TinyImageNet (alpha=0.05) | test accuracy (%): 35.02 -> 32.88 | Hurt |
| use Structured-DeCov instead of FedDecorr (ablation) | FedAvg · TinyImageNet (alpha=0.05) | test accuracy (%): 35.02 -> 32.04 | Hurt |
| add FedDecorr regularizer | FedAvg · computation time TinyImageNet | time per round (min): 25.4 -> 25.7 | No clear effect |
| vary local epochs E from 1 upward with FedDecorr (ablation) | FedAvg · CIFAR100 (alpha=0.05) | test accuracy (%): 50.67 -> 53.18 | Improved |
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)
- FedGPS: Statistical Rectification Against Data Heterogeneity in Federated Learning (NeurIPS 2025 · similarity 0.88)
- FedLoGe: Joint Local and Generic Federated Learning under Long-tailed Data (ICLR 2024 · similarity 0.88)
- FedSMU: Communication-Efficient and Generalization-Enhanced Federated Learning through Symbolic Model Updates (ICML 2025 · similarity 0.88)
- Tackling Intertwined Data and Device Heterogeneities in Federated Learning with Unlimited Staleness (AAAI 2025 · 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.
