Skip to content

WenyuMolly/ECE572_Final_Project

BranchesTags

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

77 Commits
.gitignore
.gitignore
 
 
BadNets_confusion_matrix.png
BadNets_confusion_matrix.png
 
 
BadNets_test_accuracy_curve.png
BadNets_test_accuracy_curve.png
 
 
BadNets_training_loss_curve.png
BadNets_training_loss_curve.png
 
 
README.md
README.md
 
 
TrojanNN_confusion_matrix.png
TrojanNN_confusion_matrix.png
 
 
TrojanNN_test_accuracy_curve.png
TrojanNN_test_accuracy_curve.png
 
 
TrojanNN_training_loss_curve.png
TrojanNN_training_loss_curve.png
 
 
backdoor_attack.py
backdoor_attack.py
 
 
backdoor_attacked_model_results.txt
backdoor_attacked_model_results.txt
 
 
badnets_and_trojannn.py
badnets_and_trojannn.py
 
 
badnets_model_gradcam_sample_1.png
badnets_model_gradcam_sample_1.png
 
 
badnets_model_gradcam_sample_2.png
badnets_model_gradcam_sample_2.png
 
 
badnets_model_gradcam_sample_3.png
badnets_model_gradcam_sample_3.png
 
 
badnets_model_sample_1_cam.png
badnets_model_sample_1_cam.png
 
 
badnets_model_sample_1_gradcam.png
badnets_model_sample_1_gradcam.png
 
 
badnets_model_sample_1_original.png
badnets_model_sample_1_original.png
 
 
badnets_model_sample_1_overlay.png
badnets_model_sample_1_overlay.png
 
 
badnets_model_sample_2_cam.png
badnets_model_sample_2_cam.png
 
 
badnets_model_sample_2_gradcam.png
badnets_model_sample_2_gradcam.png
 
 
badnets_model_sample_2_original.png
badnets_model_sample_2_original.png
 
 
badnets_model_sample_2_overlay.png
badnets_model_sample_2_overlay.png
 
 
badnets_model_sample_3_cam.png
badnets_model_sample_3_cam.png
 
 
badnets_model_sample_3_gradcam.png
badnets_model_sample_3_gradcam.png
 
 
badnets_model_sample_3_original.png
badnets_model_sample_3_original.png
 
 
badnets_model_sample_3_overlay.png
badnets_model_sample_3_overlay.png
 
 
clean_model_cam_sample_1.png
clean_model_cam_sample_1.png
 
 
clean_model_gradcam_sample_1.png
clean_model_gradcam_sample_1.png
 
 
clean_model_gradcam_sample_2.png
clean_model_gradcam_sample_2.png
 
 
clean_model_gradcam_sample_3.png
clean_model_gradcam_sample_3.png
 
 
clean_model_sample_1_cam.png
clean_model_sample_1_cam.png
 
 
clean_model_sample_1_gradcam.png
clean_model_sample_1_gradcam.png
 
 
clean_model_sample_1_original.png
clean_model_sample_1_original.png
 
 
clean_model_sample_1_overlay.png
clean_model_sample_1_overlay.png
 
 
clean_model_sample_2_cam.png
clean_model_sample_2_cam.png
 
 
clean_model_sample_2_gradcam.png
clean_model_sample_2_gradcam.png
 
 
clean_model_sample_2_original.png
clean_model_sample_2_original.png
 
 
clean_model_sample_2_overlay.png
clean_model_sample_2_overlay.png
 
 
clean_model_sample_3_cam.png
clean_model_sample_3_cam.png
 
 
clean_model_sample_3_gradcam.png
clean_model_sample_3_gradcam.png
 
 
clean_model_sample_3_original.png
clean_model_sample_3_original.png
 
 
clean_model_sample_3_overlay.png
clean_model_sample_3_overlay.png
 
 
clean_model_training.py
clean_model_training.py
 
 
confusion_matrix.png
confusion_matrix.png
 
 
error_info.md
error_info.md
 
 
results.txt
results.txt
 
 
results_clean_model.txt
results_clean_model.txt
 
 
test_accuracy_curve.png
test_accuracy_curve.png
 
 
training_loss_curve.png
training_loss_curve.png
 
 
trojannn_model_gradcam_sample_1.png
trojannn_model_gradcam_sample_1.png
 
 
trojannn_model_gradcam_sample_2.png
trojannn_model_gradcam_sample_2.png
 
 
trojannn_model_gradcam_sample_3.png
trojannn_model_gradcam_sample_3.png
 
 
trojannn_model_sample_1_cam.png
trojannn_model_sample_1_cam.png
 
 
trojannn_model_sample_1_gradcam.png
trojannn_model_sample_1_gradcam.png
 
 
trojannn_model_sample_1_original.png
trojannn_model_sample_1_original.png
 
 
trojannn_model_sample_1_overlay.png
trojannn_model_sample_1_overlay.png
 
 
trojannn_model_sample_2_cam.png
trojannn_model_sample_2_cam.png
 
 
trojannn_model_sample_2_gradcam.png
trojannn_model_sample_2_gradcam.png
 
 
trojannn_model_sample_2_original.png
trojannn_model_sample_2_original.png
 
 
trojannn_model_sample_2_overlay.png
trojannn_model_sample_2_overlay.png
 
 
trojannn_model_sample_3_cam.png
trojannn_model_sample_3_cam.png
 
 
trojannn_model_sample_3_gradcam.png
trojannn_model_sample_3_gradcam.png
 
 
trojannn_model_sample_3_original.png
trojannn_model_sample_3_original.png
 
 
trojannn_model_sample_3_overlay.png
trojannn_model_sample_3_overlay.png
 
 
visualization.py
visualization.py
 
 

Repository files navigation

ECE572_Final_Project - Analyzing CAM and GradCAM’s Sensitivity to Different Backdoor Attacks

Script Description

  1. clean_model_training.py This is used to train a clean resnet18 model on cifar10 dateset.

  2. badnets and trojannn.py This is used to train two backdoor attack models. One is BadNets model, which contains a static trigger, and the other one is TrojanNN model, which contains a dynamic trigger. The target class is "0".

  3. visualization.py This is used to display CAM and GradCam for each model.

About

No description, website, or topics provided.

Resources

Readme
Activity

Stars

0 stars

Watchers

1 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

No packages published

Languages