yaodongyu / TCT

TCT: Convexifying Federated Learning using Bootstrapped Neural Tangent Kernels

Train-Convexify-Train (TCT)

This is the code for the NeurIPS2022 paper [link]:

TCT: Convexifying Federated Learning using Bootstrapped Neural Tangent Kernels

by Yaodong Yu, Alexander Wei, Sai Praneeth Karimireddy, Yi Ma, Michael I. Jordan from UC Berkeley.

Prerequisites

• Python
• Pytorch
• CUDA
• numpy

* Example on MNIST

Example code on MNIST can be found under mnist folder.

Reference

For more experimental and technical details, please check our paper. If you find this useful for your work, please consider citing

@article{yu2022tct,
  title={TCT: Convexifying Federated Learning using Bootstrapped Neural Tangent Kernels},
  author={Yu, Yaodong and Wei, Alexander and Karimireddy, Sai Praneeth and Ma, Yi and Jordan, Michael I},
  journal={arXiv preprint arXiv:2207.06343},
  year={2022}
}