Self-Balancing-Federated-Learning
Self-Balancing-Federated-Learning copied to clipboard
mtang724
Implementation of the paper "Self-Balancing Federated Learning With Global Imbalanced Data in Mobile Systems"
I trained the CIFAR10 dataset on CNN, whether it's Fedavg or Astrea, with a testing accuracy of only 10 to 20, why?
Correct me if I'm wrong, but don't you need to update the kl_score if you successfully found a smaller one? 
## Beep boop. Your images are optimized! Your image file size has been reduced by **16%** 🎉 Details | File | Before | After | Percent reduction | |:--|:--|:--|:--| |...
![imgbot[bot] avatar](https://avatars.githubusercontent.com/in/4706?v=4 "imgbot[bot] avatar"){kind=avatar}