Research map / Federated learning and privacy / Communication-efficient FL
TCT: Convexifying Federated Learning using Bootstrapped Neural Tangent Kernels
Yaodong Yu, Alexander Wei, Sai Praneeth Karimireddy, Yi Ma, Michael Jordan
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 |
|---|---|---|---|
| linear probing of last layer of FedAvg-trained model | FedAvg ResNet-18 · CIFAR-10 #C=2 | test accuracy: 56.9 -> 77.9 | Improved |
| retrain last layer from FedAvg features instead of random init | ResNet-18 · CIFAR-10 | test accuracy: 35.37 -> 77.93 | Improved |
| retrain last 6 layers from FedAvg features instead of random init | ResNet-18 · CIFAR-10 | test accuracy: 91.78 -> 91.91 | No clear effect |
| replace FedAvg with TCT (train-convexify-train) | FedAvg · CIFAR-10 #C=2 | top-1 accuracy: 56.86 -> 83.02 | Improved |
| replace FedAvg with TCT | FedAvg · CIFAR-100 alpha=0.001 | top-1 accuracy: 53.89 -> 68.42 | Improved |
| replace SCAFFOLD with TCT | SCAFFOLD · CIFAR-100 alpha=0.5 | top-1 accuracy: 71.07 -> 69.68 | Hurt |
| extract eNTK from randomly initialized model (T1=0) instead of FedAvg-trained model (ablation) | TCT · CIFAR-10/100 | test accuracy | Hurt |
| use SCAFFOLD gradient correction instead of FedAvg in Stage 2 (ablation) | TCT · FMNIST, CIFAR-10, CIFAR-100 | training accuracy | Improved |
| normalize eNTK features (ablation) | TCT · CIFAR-10 #C=2 | training accuracy convergence | Improved |
| start from ImageNet pre-trained ResNet-18 instead of random bootstrapping (ablation) | TCT · CIFAR-10 #C=2 | top-1 accuracy: 83.02 -> 92.97 | Improved |
| limit Stage 2 to one communication round (ablation) | TCT · CIFAR-10 #C=2 | top-1 accuracy: 83.02 -> 64.94 | Hurt |
| use group norm instead of batch norm (ablation) | FedAvg · CIFAR-10 #C=1 | top-1 accuracy: 11.27 -> 21.23 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SCAFFOLD — SCAFFOLD: Stochastic Controlled Averaging for Federated Learning (ICML 2020) · better on 7 of 8 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- You Are Your Own Best Teacher: Achieving Centralized-level Performance in Federated Learning under Heterogeneous and Long-tailed Data (ICCV 2025 · similarity 0.88)
- Federated Learning for Feature Generalization with Convex Constraints (ICML 2025 · similarity 0.88)
- NTK-DFL: Enhancing Decentralized Federated Learning in Heterogeneous Settings via Neural Tangent Kernel (ICML 2025 · similarity 0.87)
- FedQS: Optimizing Gradient and Model Aggregation for Semi-Asynchronous Federated Learning (NeurIPS 2025 · similarity 0.87)
- Improved Coresets for Vertical Federated Learning: Regularized Linear and Logistic Regressions (ICML 2025 · similarity 0.87)
- Widening the Network Mitigates the Impact of Data Heterogeneity on FedAvg (ICML 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.
