Boyu Chen

FedDKD: Federated Learning with Decentralized Knowledge Distillation

2022 · arXiv (Cornell University)

The performance of federated learning in neural networks is generally influenced by the heterogeneity of the data distribution. For a well-performing global model, taking a weighted average of the local models, as done by most …

Modify Training Directions in Function Space to Reduce Generalization Error

2023 · arXiv (Cornell University)

We propose theoretical analyses of a modified natural gradient descent method in the neural network function space based on the eigendecompositions of neural tangent kernel and Fisher information matrix. We firstly present analytical expression for …