search

skylorh / FedCLASS

MIT License
18 stars 3 forks source link

Inconsistent results ？ #1

Open Eurekaer opened 8 months ago

Eurekaer commented 8 months ago

Hello, thank you very much for contributing such a good job. I have some doubts when executing the code you provided, and I hope you can answer them. Thank you very much.

  1. First, I changed the model_name in ce.json to ours, dataset to cifar10, and the result is as follows: image It seems to be somewhat different from the results mentioned in your paper. I don't quite understand it. I hope I can get your explanation
  2. Secondly, there is CENet in your code, which doesn't seem to be mentioned in the paper. Is this a new attempt you made?
  3. Finally, you mentioned the results of FedAvg in your paper and compared them with federated incremental learning methods, including your proposed method. In this case, how does it work? I don't seem to see this part in the code. Thanks again!
wuzhiyuan2000 commented 4 months ago

I've responded your email. Thanks for your questions