Steve Chien comments

Repositories
Issues
Comments

Results 4 comments of


                                            Steve Chien

AssertionError: compute_gradients() on the differentially private optimizer was not called. Which means that the training is not differentially private. It happens for example in Keras training in TensorFlow 2.0+.

Hi @Flyige and @mcps5601 ... Sorry for the delay in replying. The DP-SGD optimizers work by clipping the per-microbatch gradients and adding noise. This is done in the method `compute_gradients`....

The accuracy of DPSGD_Kears only have 10%

Thanks for the report-- I'll take a look.

The accuracy of DPSGD_Kears only have 10%

Hi @Mrxulovemingming , Can you share your command line? I just tried this command line: `python3 code/tutorials/mnist_dpsgd_tutorial_keras.py` with no extra arguments and got this output: ![Screen Shot 2021-04-12 at 10...

The accuracy of DPSGD_Kears only have 10%

Thanks for the extra information. I'm looking into it. By the way, is there an image missing in your last message?