Skip to content

Commit b331b3a

Browse files
committed
Update Readme.md
1 parent eaf6c58 commit b331b3a

File tree

1 file changed

+37
-27
lines changed

1 file changed

+37
-27
lines changed

Readme.md

Lines changed: 37 additions & 27 deletions
Original file line numberDiff line numberDiff line change
@@ -1,26 +1,38 @@
11
# Attack code
22

3-
## The composition of the directoty
4-
5-
- 0_AEEA_dataset
6-
Our datasets are here including a log dataset,a traffic dataset.
7-
- 1_model_for_traffic
8-
This directory includes our pretrained models in traffic dataset. You can get the training process by the tensorboard.
9-
- 2_model_for_log
10-
This directory includes our pretrained models in traffic dataset. You can get the training process by the tensorboard.
11-
- 3_attack_code
12-
You can get all the training code in this directory.
13-
If you want to test your own model, you can add your model in the your_model_name.py and put your pretrained model here.
14-
You can also try different ways to attck models, such as random attack,differential evolution.
15-
It is convinient to try your models in different dataset.
16-
- 4_EVALUATION
17-
- 5_For_TEST_h5
18-
You can get the accurate attack resutls.
19-
20-
## How to attack your models
21-
22-
- You need to pass your args to attack_for_traffic.py to attack models. EN:You need to try your own models before you attack it.
23-
- Example: python model_name.py --model model_name --others
3+
## What can I do with these files?
4+
5+
1. You can use attack_for_AE/log/traffic.py to attack different datasets with different models in DE.
6+
2. You can find different models in the DFR/happy/lenet/DFR_log/happy_log/lenet_log.py.
7+
3. Meanwhile you can randomly attactk models by random_arrack/random_generate/random_generate_for_log.py.
8+
9+
## where can I get the pretrained models and datasets?
10+
11+
1. [url](https://pan.baidu.com/s/1z6F8n5GpKqA2yjRtY-Uojw)
12+
password:6mw8
13+
14+
2. The composition of the directoty in the url:
15+
- 0_AEEA_dataset
16+
Our datasets are here including a log dataset,a traffic dataset.
17+
- 1_model_for_traffic
18+
This directory includes our pretrained models in traffic dataset. You can get the training process by the tensorboard.
19+
- 2_model_for_log
20+
This directory includes our pretrained models in traffic dataset. You can get the training process by the tensorboard.
21+
- 3_attack_code
22+
You can get all the training code in this directory.
23+
If you want to test your own model, you can add your model in the your_model_name.py and put your pretrained model here.
24+
You can also try different ways to attck models, such as random attack,differential evolution.
25+
It is convinient to try your models in different dataset.
26+
- 4_EVALUATION
27+
- 5_For_TEST_h5
28+
You can get the accurate attack resutls.
29+
30+
## How can I attack my models?
31+
32+
- You need to pass your args to attack_for_traffic.py(or other attack files) to attack models.
33+
>EN:You need to train your own models before you attack it.
34+
- Example: python model_name.py --model model_name --other_args
35+
- To get more args, you can read attack_for_traffic.py.
2436

2537
## Envirionment
2638

@@ -29,9 +41,7 @@ keras
2941

3042
## Some important tips
3143

32-
- You can write your own model that you want to attack in keras. And you need to follow the rules in the 3_attack_code/model_name.py.
33-
- I write some comments for attacking models in the 3_attack_code/attack_for_traffic.py.
34-
- I write some comments for model building in the 3_attack_code/happy.py.
35-
- You can find the specific implements of differential_evolution in the 3_attack_code/differential_evolution.py.
36-
37-
## GOOD LUCKY TO YOUR TRAVEL IN AI!
44+
- You can write your own model that you want to attack in keras. And you need to follow the examples in the happy.py.
45+
- I write some comments for attacking models in the attack_for_traffic.py.
46+
- I write some comments for model building in the happy.py.
47+
- You can find the specific implements of differential_evolution in the differential_evolution.py.

0 commit comments

Comments
 (0)