diogenes0319 / fedmd_clean

FedMD: Heterogenous Federated Learning via Model Distillation

FedMD

FedMD: Heterogenous Federated Learning via Model Distillation. Preprint on https://arxiv.org/abs/1910.03581.

Run scripts on Google Colab

1. open a google Colab

2. Clone the project folder from Github

! git clone github_link

3. Then access the folder just created.

% cd project_folder/

4. Run the python script in Colab. For instance

! python FEMNIST_Balanced.py -conf conf/EMNIST_balance_conf.json

fedmd_clean's Issues

Diagram of simulation results

Hi, there! May I ask that where can I see the program result chart?Are the pictures automatically generated after the program has been run?If the program is not finished, the result graph can not be viewed?

Can't find a file indicates package versions. Could you provide a .yaml file or something similar? Thank you~

The test accuracy achieve highest at digest phase, and decreased at revisit phase.

Hi,
When I use this code to do the experiment on CIFAR10, I find that the test accuracy achieve highest at digest phase, and decreased greatly at revisit phase, is this normal？Should the clients take the model after the digest phase instead of the revisit phase?

ValueError: Layer model expects 1 input(s), but it received 2 input tensors.

Applicability of FedMD for regression tasks

Hi Junpu,

Thank you for the interesting paper and the public code. I understand that FedMD works on classification tasks. May I check if it is also applicable for time series regression modelling?

Thank you!

Hi! It seems that the code is not complete.

Or the code is complete, but it is not reflected in the readme.md?

ValueError: Layer model expects 1 input(s), but it received 2 input tensors

Hello, I have encountered the following problem but failed to find a solution. Could you have a look？
“ValueError: Layer model expects 1 input(s), but it received 2 input tensors. Inputs received: [<tf.Tensor 'IteratorGetNext:0' shape=(None, 32, 32, 3) dtype=float32>, <tf.Tensor 'ExpandDims:0' shape=(None, 1) dtype=uint8>]”