Paper comparisons
オープン
documentation
priority - 4
- 主要言語
- Python
- スター
- 23
- フォーク
- 28
- 平均マージ
- 4日 12時間
- マージ済み PR（30日）
- 5
説明
# Issue
## Problem Description
So far, the implemented attacks have been verified on CIFAR-10 using small convolutional models. We need to ensure that the attacks are true to their origins.
## Expected Behavior
Attacks should resemble the performance of the papers.
## What Needs to be Done
Test the attacks on a representative case from the papers.
Proposal: targer model: ResNet18, dataset: CINIC (https://github.com/BayesWatch/cinic-10)
## How Can It Be Tested or Reproduced
Run each of the attacks and document the result
コントリビューションガイド
評価
この issue はまだ評価されていません。