search

sutd-visual-computing-group / Re-thinking_MI

[CVPR-2023] Re-thinking Model Inversion Attacks Against Deep Neural Networks
https://ngoc-nguyen-0.github.io/re-thinking_model_inversion_attacks/
32 stars 4 forks source link

The acc of the provided checkpoint is very low #1

Closed ghost closed 1 year ago

ghost commented 1 year ago

Hello! I used the checkpoints of the target classifier VGG16 and IR152 provided by you and tested the classification accuracy. Why is the acc only 9.97%?

ngoc-nguyen-0 commented 1 year ago

Hi,

To ensure accurate evaluation, I recommend testing your evaluation code with the pretrained model provided by KEDMI author. You can find the model at the following GitHub repository: https://github.com/SCccc21/Knowledge-Enriched-DMI

If you encounter low classification accuracy, it's possible that there might be some issues in your evaluation code, such as the dataloader. I suggest double-checking your implementation for any bugs or errors.

Ngoc