Tinsae/Finetuning
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

attcks

Browse Source
This commit is contained in:
Tinsae
2026-07-08 10:07:39 +02:00
parent eb8060fc05
commit a118a975c1
20 changed files with 287 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
Tune_new.py
View File

@@ -290,15 +290,15 @@ def run_unlearning_and_strategy_eval(env_dict, forget_class_idx, strategy, evalu
base_shadow_instance=shadow_model, base_shadow_instance=shadow_model,
device=device device=device
) )
else:
# Define validation tracking steps dynamically # Define validation tracking steps dynamically
evaluation_domains = [ evaluation_domains = [
{"loader": retain_test_loader, "mode": "retain", "label": "\n--- Performance on Retained Classes"}, {"loader": retain_test_loader, "mode": "retain", "label": "\n--- Performance on Retained Classes"},
{"loader": forget_test_loader, "mode": "forget", "label": "\n--- Performance on Forgotten Class"}, {"loader": forget_test_loader, "mode": "forget", "label": "\n--- Performance on Forgotten Class"},
{"loader": forget_train_loader, "mode": "forget_train", "label": "\n--- Performance on Forgotten Class (Train Set - Verifying Unlearning)"} {"loader": forget_train_loader, "mode": "forget_train", "label": "\n--- Performance on Forgotten Class (Train Set - Verifying Unlearning)"}
] ]
log_metrics(evaluation_domains, reloaded, strategy_in_use) log_metrics(evaluation_domains, reloaded, strategy_in_use)
# entry # entry
@@ -384,7 +384,7 @@ if __name__ == "__main__":
strategy=strategy, strategy=strategy,
# if we are finetuning, no need to evaluate base model. # if we are finetuning, no need to evaluate base model.
# or may be never when not either! # or may be never when not either!
evaluate = False, evaluate = not finetuning,
suite_runner=suite_runner suite_runner=suite_runner
) )
# just a single class run before running all remaining classes. # just a single class run before running all remaining classes.

13
architectures/Model.py
View File

@@ -24,18 +24,7 @@ class Model(ABC):
pass pass
''' '''
Have to have a new param here as mode, for example it would be base, or retrain
param mode = "base" or "retrain"
that way I can save time it takes to train and retrain.
file would be solved with Util functions
log_file = Path(f"reports/{mode}/{self.__class__.__name__}/time_metrics.txt")
Util._initialize_log_file(log_file):
strt = time.perf_counter()
end = time.perf_counter()
and then save logs
execution_time = end -strt
Util.log_metric(log_file, execution_time: float):
''' '''
def train(self, epochs, loader, rate, mode = "retrain"): def train(self, epochs, loader, rate, mode = "retrain"):
criterion = nn.CrossEntropyLoss() criterion = nn.CrossEntropyLoss()

12
reports/CertifiedUnlearning/RESNET34/forget.csv
View File

@@ -364,3 +364,15 @@ accuracy,macro_precision,macro_recall,macro_f1,weighted_precision,weighted_recal
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0125,1.0000,0.0125,0.0247,1.0000,0.0125,0.0247 0.0125,1.0000,0.0125,0.0247,1.0000,0.0125,0.0247
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.4375,1.0000,0.4375,0.6087,1.0000,0.4375,0.6087
0.3875,1.0000,0.3875,0.5586,1.0000,0.3875,0.5586
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
1 accuracy macro_precision macro_recall macro_f1 weighted_precision weighted_recall weighted_f1
364 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
365 0.0125 1.0000 0.0125 0.0247 1.0000 0.0125 0.0247
366 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
367 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
368 0.4375 1.0000 0.4375 0.6087 1.0000 0.4375 0.6087
369 0.3875 1.0000 0.3875 0.5586 1.0000 0.3875 0.5586
370 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
371 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
372 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
373 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
374 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
375 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
376 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
377 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
378 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

12
reports/CertifiedUnlearning/RESNET34/forget_train.csv
View File

@@ -364,3 +364,15 @@ accuracy,macro_precision,macro_recall,macro_f1,weighted_precision,weighted_recal
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.5062,1.0000,0.5062,0.6722,1.0000,0.5062,0.6722
0.4750,1.0000,0.4750,0.6441,1.0000,0.4750,0.6441
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
1 accuracy macro_precision macro_recall macro_f1 weighted_precision weighted_recall weighted_f1
364 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
365 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
366 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
367 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
368 0.5062 1.0000 0.5062 0.6722 1.0000 0.5062 0.6722
369 0.4750 1.0000 0.4750 0.6441 1.0000 0.4750 0.6441
370 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
371 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
372 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
373 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
374 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
375 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
376 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
377 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
378 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

12
reports/CertifiedUnlearning/RESNET34/retain.csv
View File

@@ -364,3 +364,15 @@ accuracy,macro_precision,macro_recall,macro_f1,weighted_precision,weighted_recal
0.8730,0.9165,0.8730,0.8827,0.9165,0.8730,0.8827 0.8730,0.9165,0.8730,0.8827,0.9165,0.8730,0.8827
0.9454,0.9470,0.9454,0.9454,0.9470,0.9454,0.9454 0.9454,0.9470,0.9454,0.9454,0.9470,0.9454,0.9454
0.5987,0.7725,0.5987,0.5938,0.7725,0.5987,0.5938 0.5987,0.7725,0.5987,0.5938,0.7725,0.5987,0.5938
0.1961,0.3286,0.1961,0.1473,0.3286,0.1961,0.1473
0.9474,0.9490,0.9474,0.9475,0.9490,0.9474,0.9475
0.9461,0.9485,0.9461,0.9462,0.9485,0.9461,0.9462
0.6309,0.8254,0.6309,0.6278,0.8254,0.6309,0.6278
0.9079,0.9297,0.9079,0.9121,0.9297,0.9079,0.9121
0.0526,0.0028,0.0526,0.0053,0.0028,0.0526,0.0053
0.8928,0.9239,0.8928,0.8992,0.9239,0.8928,0.8992
0.9316,0.9375,0.9316,0.9323,0.9375,0.9316,0.9323
0.8822,0.9078,0.8822,0.8844,0.9078,0.8822,0.8844
0.9053,0.9214,0.9053,0.9081,0.9214,0.9053,0.9081
0.0533,0.0343,0.0533,0.0077,0.0343,0.0533,0.0077
0.7171,0.8599,0.7171,0.7316,0.8599,0.7171,0.7316
1 accuracy macro_precision macro_recall macro_f1 weighted_precision weighted_recall weighted_f1
364 0.8730 0.9165 0.8730 0.8827 0.9165 0.8730 0.8827
365 0.9454 0.9470 0.9454 0.9454 0.9470 0.9454 0.9454
366 0.5987 0.7725 0.5987 0.5938 0.7725 0.5987 0.5938
367 0.1961 0.3286 0.1961 0.1473 0.3286 0.1961 0.1473
368 0.9474 0.9490 0.9474 0.9475 0.9490 0.9474 0.9475
369 0.9461 0.9485 0.9461 0.9462 0.9485 0.9461 0.9462
370 0.6309 0.8254 0.6309 0.6278 0.8254 0.6309 0.6278
371 0.9079 0.9297 0.9079 0.9121 0.9297 0.9079 0.9121
372 0.0526 0.0028 0.0526 0.0053 0.0028 0.0526 0.0053
373 0.8928 0.9239 0.8928 0.8992 0.9239 0.8928 0.8992
374 0.9316 0.9375 0.9316 0.9323 0.9375 0.9316 0.9323
375 0.8822 0.9078 0.8822 0.8844 0.9078 0.8822 0.8844
376 0.9053 0.9214 0.9053 0.9081 0.9214 0.9053 0.9081
377 0.0533 0.0343 0.0533 0.0077 0.0343 0.0533 0.0077
378 0.7171 0.8599 0.7171 0.7316 0.8599 0.7171 0.7316

29
reports/CertifiedUnlearning/attack_values.csv
View File

@@ -2,3 +2,32 @@ target_class,parameter_mia_accuracy,latent_distance_tell,lookalike_accuracy
0,0.500000,7.219862,0.979167 0,0.500000,7.219862,0.979167
1,0.500000,3.659238,0.958333 1,0.500000,3.659238,0.958333
2,0.500000,6.939345,0.885417 2,0.500000,6.939345,0.885417
3,0.500000,7.800516,0.989583
4,0.500000,6.631370,1.000000
5,0.500000,7.557893,0.937500
6,0.500000,8.085600,0.937500
7,0.500000,7.518537,0.947917
8,0.500000,9.040818,0.895833
9,0.500000,8.269253,0.968750
10,0.500000,5.678658,1.000000
11,0.450000,6.800340,1.000000
12,0.500000,7.708953,0.937500
13,0.500000,6.559650,1.000000
14,0.500000,8.420962,0.989583
15,0.500000,7.526255,1.000000
16,0.500000,6.906671,0.968750
17,0.500000,5.499599,0.979167
18,0.500000,8.465704,0.979167
19,0.500000,5.958429,0.979167
0,0.500000,5.903083,0.979167
1,0.500000,4.479048,0.947917
2,0.500000,4.460630,0.979167
3,0.500000,9.916990,1.000000
4,0.500000,4.245532,0.979167
5,0.500000,5.771674,1.000000
6,0.500000,6.300947,0.979167
7,0.500000,6.803962,0.979167
8,0.500000,9.174347,1.000000
9,0.500000,7.843153,0.989583
10,0.500000,5.029855,1.000000
11,0.500000,7.945799,1.000000
1 target_class parameter_mia_accuracy latent_distance_tell lookalike_accuracy
2 0 0.500000 7.219862 0.979167
3 1 0.500000 3.659238 0.958333
4 2 0.500000 6.939345 0.885417
5 3 0.500000 7.800516 0.989583
6 4 0.500000 6.631370 1.000000
7 5 0.500000 7.557893 0.937500
8 6 0.500000 8.085600 0.937500
9 7 0.500000 7.518537 0.947917
10 8 0.500000 9.040818 0.895833
11 9 0.500000 8.269253 0.968750
12 10 0.500000 5.678658 1.000000
13 11 0.450000 6.800340 1.000000
14 12 0.500000 7.708953 0.937500
15 13 0.500000 6.559650 1.000000
16 14 0.500000 8.420962 0.989583
17 15 0.500000 7.526255 1.000000
18 16 0.500000 6.906671 0.968750
19 17 0.500000 5.499599 0.979167
20 18 0.500000 8.465704 0.979167
21 19 0.500000 5.958429 0.979167
22 0 0.500000 5.903083 0.979167
23 1 0.500000 4.479048 0.947917
24 2 0.500000 4.460630 0.979167
25 3 0.500000 9.916990 1.000000
26 4 0.500000 4.245532 0.979167
27 5 0.500000 5.771674 1.000000
28 6 0.500000 6.300947 0.979167
29 7 0.500000 6.803962 0.979167
30 8 0.500000 9.174347 1.000000
31 9 0.500000 7.843153 0.989583
32 10 0.500000 5.029855 1.000000
33 11 0.500000 7.945799 1.000000

14
reports/LinearFiltration/RESNET34/forget.csv
View File

@@ -436,3 +436,17 @@ accuracy,macro_precision,macro_recall,macro_f1,weighted_precision,weighted_recal
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
1 accuracy macro_precision macro_recall macro_f1 weighted_precision weighted_recall weighted_f1
436 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
437 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
438 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
439 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
440 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
441 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
442 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
443 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
444 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
445 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
446 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
447 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
448 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
449 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
450 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
451 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
452 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

14
reports/LinearFiltration/RESNET34/forget_train.csv
View File

@@ -436,3 +436,17 @@ accuracy,macro_precision,macro_recall,macro_f1,weighted_precision,weighted_recal
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
1 accuracy macro_precision macro_recall macro_f1 weighted_precision weighted_recall weighted_f1
436 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
437 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
438 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
439 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
440 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
441 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
442 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
443 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
444 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
445 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
446 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
447 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
448 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
449 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
450 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
451 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
452 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

14
reports/LinearFiltration/RESNET34/retain.csv
View File

@@ -436,3 +436,17 @@ accuracy,macro_precision,macro_recall,macro_f1,weighted_precision,weighted_recal
0.9546,0.9556,0.9546,0.9547,0.9556,0.9546,0.9547 0.9546,0.9556,0.9546,0.9547,0.9556,0.9546,0.9547
0.9533,0.9543,0.9533,0.9534,0.9543,0.9533,0.9534 0.9533,0.9543,0.9533,0.9534,0.9543,0.9533,0.9534
0.9533,0.9543,0.9533,0.9534,0.9543,0.9533,0.9534 0.9533,0.9543,0.9533,0.9534,0.9543,0.9533,0.9534
0.9526,0.9541,0.9526,0.9529,0.9541,0.9526,0.9529
0.9526,0.9541,0.9526,0.9529,0.9541,0.9526,0.9529
0.9533,0.9551,0.9533,0.9536,0.9551,0.9533,0.9536
0.9533,0.9545,0.9533,0.9535,0.9545,0.9533,0.9535
0.9546,0.9562,0.9546,0.9548,0.9562,0.9546,0.9548
0.9553,0.9570,0.9553,0.9555,0.9570,0.9553,0.9555
0.9566,0.9581,0.9566,0.9568,0.9581,0.9566,0.9568
0.9553,0.9568,0.9553,0.9555,0.9568,0.9553,0.9555
0.9513,0.9532,0.9513,0.9516,0.9532,0.9513,0.9516
0.9539,0.9556,0.9539,0.9542,0.9556,0.9539,0.9542
0.9520,0.9539,0.9520,0.9523,0.9539,0.9520,0.9523
0.9533,0.9547,0.9533,0.9535,0.9547,0.9533,0.9535
0.9546,0.9561,0.9546,0.9549,0.9561,0.9546,0.9549
0.9513,0.9531,0.9513,0.9516,0.9531,0.9513,0.9516
1 accuracy macro_precision macro_recall macro_f1 weighted_precision weighted_recall weighted_f1
436 0.9546 0.9556 0.9546 0.9547 0.9556 0.9546 0.9547
437 0.9533 0.9543 0.9533 0.9534 0.9543 0.9533 0.9534
438 0.9533 0.9543 0.9533 0.9534 0.9543 0.9533 0.9534
439 0.9526 0.9541 0.9526 0.9529 0.9541 0.9526 0.9529
440 0.9526 0.9541 0.9526 0.9529 0.9541 0.9526 0.9529
441 0.9533 0.9551 0.9533 0.9536 0.9551 0.9533 0.9536
442 0.9533 0.9545 0.9533 0.9535 0.9545 0.9533 0.9535
443 0.9546 0.9562 0.9546 0.9548 0.9562 0.9546 0.9548
444 0.9553 0.9570 0.9553 0.9555 0.9570 0.9553 0.9555
445 0.9566 0.9581 0.9566 0.9568 0.9581 0.9566 0.9568
446 0.9553 0.9568 0.9553 0.9555 0.9568 0.9553 0.9555
447 0.9513 0.9532 0.9513 0.9516 0.9532 0.9513 0.9516
448 0.9539 0.9556 0.9539 0.9542 0.9556 0.9539 0.9542
449 0.9520 0.9539 0.9520 0.9523 0.9539 0.9520 0.9523
450 0.9533 0.9547 0.9533 0.9535 0.9547 0.9533 0.9535
451 0.9546 0.9561 0.9546 0.9549 0.9561 0.9546 0.9549
452 0.9513 0.9531 0.9513 0.9516 0.9531 0.9513 0.9516

30
reports/LinearFiltration/attack_values.csv
View File

@@ -3,3 +3,33 @@ target_class,parameter_mia_accuracy,latent_distance_tell,lookalike_accuracy
1,0.500000,3.573733,1.000000 1,0.500000,3.573733,1.000000
2,0.500000,3.924550,1.000000 2,0.500000,3.924550,1.000000
3,0.500000,3.515182,1.000000 3,0.500000,3.515182,1.000000
4,0.500000,3.369485,1.000000
5,0.500000,4.697085,1.000000
6,0.500000,4.185269,1.000000
7,0.500000,3.473583,1.000000
8,0.500000,4.518519,1.000000
9,0.500000,4.484542,1.000000
10,0.500000,3.353875,1.000000
11,0.500000,3.815037,0.989583
12,0.500000,3.493994,1.000000
13,0.500000,4.018041,1.000000
14,0.500000,4.059689,1.000000
15,0.500000,4.097189,1.000000
16,0.500000,3.287521,1.000000
17,0.500000,3.872488,1.000000
18,0.500000,3.698952,0.989583
19,0.500000,3.844830,0.989583
0,0.500000,3.413494,1.000000
0,0.500000,3.416399,1.000000
1,0.500000,3.396268,1.000000
2,0.500000,3.648122,1.000000
3,0.500000,3.889339,0.989583
4,0.500000,3.408539,1.000000
5,0.500000,4.625342,1.000000
6,0.500000,3.925502,0.989583
7,0.500000,3.033561,0.989583
8,0.500000,3.819170,0.989583
9,0.500000,4.497495,1.000000
10,0.500000,3.701839,1.000000
11,0.500000,4.369391,0.989583
12,0.500000,3.945474,1.000000
1 target_class parameter_mia_accuracy latent_distance_tell lookalike_accuracy
3 1 0.500000 3.573733 1.000000
4 2 0.500000 3.924550 1.000000
5 3 0.500000 3.515182 1.000000
6 4 0.500000 3.369485 1.000000
7 5 0.500000 4.697085 1.000000
8 6 0.500000 4.185269 1.000000
9 7 0.500000 3.473583 1.000000
10 8 0.500000 4.518519 1.000000
11 9 0.500000 4.484542 1.000000
12 10 0.500000 3.353875 1.000000
13 11 0.500000 3.815037 0.989583
14 12 0.500000 3.493994 1.000000
15 13 0.500000 4.018041 1.000000
16 14 0.500000 4.059689 1.000000
17 15 0.500000 4.097189 1.000000
18 16 0.500000 3.287521 1.000000
19 17 0.500000 3.872488 1.000000
20 18 0.500000 3.698952 0.989583
21 19 0.500000 3.844830 0.989583
22 0 0.500000 3.413494 1.000000
23 0 0.500000 3.416399 1.000000
24 1 0.500000 3.396268 1.000000
25 2 0.500000 3.648122 1.000000
26 3 0.500000 3.889339 0.989583
27 4 0.500000 3.408539 1.000000
28 5 0.500000 4.625342 1.000000
29 6 0.500000 3.925502 0.989583
30 7 0.500000 3.033561 0.989583
31 8 0.500000 3.819170 0.989583
32 9 0.500000 4.497495 1.000000
33 10 0.500000 3.701839 1.000000
34 11 0.500000 4.369391 0.989583
35 12 0.500000 3.945474 1.000000

14
reports/Retrain/RESNET34/forget.csv
View File

@@ -66,3 +66,17 @@ accuracy,macro_precision,macro_recall,macro_f1,weighted_precision,weighted_recal
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
1 accuracy macro_precision macro_recall macro_f1 weighted_precision weighted_recall weighted_f1
66 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
67 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
68 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
69 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
70 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
71 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
72 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
73 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
74 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
75 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
76 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
77 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
78 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
79 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
80 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
81 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
82 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

14
reports/Retrain/RESNET34/forget_train.csv
View File

@@ -66,3 +66,17 @@ accuracy,macro_precision,macro_recall,macro_f1,weighted_precision,weighted_recal
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
1 accuracy macro_precision macro_recall macro_f1 weighted_precision weighted_recall weighted_f1
66 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
67 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
68 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
69 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
70 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
71 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
72 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
73 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
74 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
75 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
76 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
77 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
78 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
79 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
80 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
81 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
82 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

14
reports/Retrain/RESNET34/retain.csv
View File

@@ -66,3 +66,17 @@ accuracy,macro_precision,macro_recall,macro_f1,weighted_precision,weighted_recal
0.9533,0.9543,0.9533,0.9534,0.9543,0.9533,0.9534 0.9533,0.9543,0.9533,0.9534,0.9543,0.9533,0.9534
0.9586,0.9599,0.9586,0.9588,0.9599,0.9586,0.9588 0.9586,0.9599,0.9586,0.9588,0.9599,0.9586,0.9588
0.9612,0.9620,0.9612,0.9613,0.9620,0.9612,0.9613 0.9612,0.9620,0.9612,0.9613,0.9620,0.9612,0.9613
0.9559,0.9571,0.9559,0.9560,0.9571,0.9559,0.9560
0.9559,0.9571,0.9559,0.9560,0.9571,0.9559,0.9560
0.9507,0.9518,0.9507,0.9507,0.9518,0.9507,0.9507
0.9612,0.9622,0.9612,0.9612,0.9622,0.9612,0.9612
0.9553,0.9561,0.9553,0.9552,0.9561,0.9553,0.9552
0.9618,0.9626,0.9618,0.9618,0.9626,0.9618,0.9618
0.9566,0.9586,0.9566,0.9569,0.9586,0.9566,0.9569
0.9533,0.9548,0.9533,0.9535,0.9548,0.9533,0.9535
0.9539,0.9548,0.9539,0.9541,0.9548,0.9539,0.9541
0.9500,0.9511,0.9500,0.9500,0.9511,0.9500,0.9500
0.9566,0.9572,0.9566,0.9565,0.9572,0.9566,0.9565
0.9546,0.9554,0.9546,0.9546,0.9554,0.9546,0.9546
0.9612,0.9620,0.9612,0.9613,0.9620,0.9612,0.9613
0.9539,0.9551,0.9539,0.9540,0.9551,0.9539,0.9540
1 accuracy macro_precision macro_recall macro_f1 weighted_precision weighted_recall weighted_f1
66 0.9533 0.9543 0.9533 0.9534 0.9543 0.9533 0.9534
67 0.9586 0.9599 0.9586 0.9588 0.9599 0.9586 0.9588
68 0.9612 0.9620 0.9612 0.9613 0.9620 0.9612 0.9613
69 0.9559 0.9571 0.9559 0.9560 0.9571 0.9559 0.9560
70 0.9559 0.9571 0.9559 0.9560 0.9571 0.9559 0.9560
71 0.9507 0.9518 0.9507 0.9507 0.9518 0.9507 0.9507
72 0.9612 0.9622 0.9612 0.9612 0.9622 0.9612 0.9612
73 0.9553 0.9561 0.9553 0.9552 0.9561 0.9553 0.9552
74 0.9618 0.9626 0.9618 0.9618 0.9626 0.9618 0.9618
75 0.9566 0.9586 0.9566 0.9569 0.9586 0.9566 0.9569
76 0.9533 0.9548 0.9533 0.9535 0.9548 0.9533 0.9535
77 0.9539 0.9548 0.9539 0.9541 0.9548 0.9539 0.9541
78 0.9500 0.9511 0.9500 0.9500 0.9511 0.9500 0.9500
79 0.9566 0.9572 0.9566 0.9565 0.9572 0.9566 0.9565
80 0.9546 0.9554 0.9546 0.9546 0.9554 0.9546 0.9546
81 0.9612 0.9620 0.9612 0.9613 0.9620 0.9612 0.9613
82 0.9539 0.9551 0.9539 0.9540 0.9551 0.9539 0.9540

30
reports/Retrain/attack_values.csv
View File

@@ -3,3 +3,33 @@ target_class,parameter_mia_accuracy,latent_distance_tell,lookalike_accuracy
1,0.500000,12.793120,0.000000 1,0.500000,12.793120,0.000000
2,0.500000,12.951434,0.000000 2,0.500000,12.951434,0.000000
3,0.500000,10.942259,0.000000 3,0.500000,10.942259,0.000000
4,0.500000,10.675704,0.000000
5,0.500000,13.046491,0.000000
6,0.500000,11.343709,0.000000
7,0.500000,10.020437,0.000000
8,0.500000,12.896643,0.000000
9,0.500000,12.220088,0.000000
10,0.500000,11.315407,0.000000
11,0.450000,10.556920,0.000000
12,0.500000,10.925443,0.000000
13,0.500000,12.007375,0.000000
14,0.500000,11.471087,0.000000
15,0.500000,12.301955,0.000000
16,0.500000,10.831913,0.000000
17,0.500000,10.942777,0.000000
18,0.500000,11.699934,0.000000
19,0.500000,12.526654,0.000000
0,0.500000,11.601441,0.000000
0,0.500000,11.639872,0.000000
1,0.500000,12.723348,0.000000
2,0.500000,13.151769,0.000000
3,0.500000,10.969514,0.000000
4,0.500000,10.674448,0.000000
5,0.500000,13.068334,0.000000
6,0.500000,11.328772,0.000000
7,0.500000,9.926997,0.000000
8,0.500000,12.859781,0.000000
9,0.500000,12.061253,0.000000
10,0.500000,11.358070,0.000000
11,0.500000,10.484346,0.000000
12,0.500000,10.926878,0.000000
1 target_class parameter_mia_accuracy latent_distance_tell lookalike_accuracy
3 1 0.500000 12.793120 0.000000
4 2 0.500000 12.951434 0.000000
5 3 0.500000 10.942259 0.000000
6 4 0.500000 10.675704 0.000000
7 5 0.500000 13.046491 0.000000
8 6 0.500000 11.343709 0.000000
9 7 0.500000 10.020437 0.000000
10 8 0.500000 12.896643 0.000000
11 9 0.500000 12.220088 0.000000
12 10 0.500000 11.315407 0.000000
13 11 0.450000 10.556920 0.000000
14 12 0.500000 10.925443 0.000000
15 13 0.500000 12.007375 0.000000
16 14 0.500000 11.471087 0.000000
17 15 0.500000 12.301955 0.000000
18 16 0.500000 10.831913 0.000000
19 17 0.500000 10.942777 0.000000
20 18 0.500000 11.699934 0.000000
21 19 0.500000 12.526654 0.000000
22 0 0.500000 11.601441 0.000000
23 0 0.500000 11.639872 0.000000
24 1 0.500000 12.723348 0.000000
25 2 0.500000 13.151769 0.000000
26 3 0.500000 10.969514 0.000000
27 4 0.500000 10.674448 0.000000
28 5 0.500000 13.068334 0.000000
29 6 0.500000 11.328772 0.000000
30 7 0.500000 9.926997 0.000000
31 8 0.500000 12.859781 0.000000
32 9 0.500000 12.061253 0.000000
33 10 0.500000 11.358070 0.000000
34 11 0.500000 10.484346 0.000000
35 12 0.500000 10.926878 0.000000

12
reports/WeightFiltration/RESNET34/forget.csv
View File

@@ -527,3 +527,15 @@ accuracy,macro_precision,macro_recall,macro_f1,weighted_precision,weighted_recal
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0375,1.0000,0.0375,0.0723,1.0000,0.0375,0.0723
0.1750,1.0000,0.1750,0.2979,1.0000,0.1750,0.2979
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0375,1.0000,0.0375,0.0723,1.0000,0.0375,0.0723
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
1 accuracy macro_precision macro_recall macro_f1 weighted_precision weighted_recall weighted_f1
527 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
528 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
529 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
530 0.0375 1.0000 0.0375 0.0723 1.0000 0.0375 0.0723
531 0.1750 1.0000 0.1750 0.2979 1.0000 0.1750 0.2979
532 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
533 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
534 0.0375 1.0000 0.0375 0.0723 1.0000 0.0375 0.0723
535 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
536 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
537 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
538 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
539 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
540 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
541 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

12
reports/WeightFiltration/RESNET34/forget_train.csv
View File

@@ -527,3 +527,15 @@ accuracy,macro_precision,macro_recall,macro_f1,weighted_precision,weighted_recal
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000 0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0375,1.0000,0.0375,0.0723,1.0000,0.0375,0.0723
0.2281,1.0000,0.2281,0.3715,1.0000,0.2281,0.3715
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0094,1.0000,0.0094,0.0186,1.0000,0.0094,0.0186
0.0094,1.0000,0.0094,0.0186,1.0000,0.0094,0.0186
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
0.0063,1.0000,0.0063,0.0124,1.0000,0.0063,0.0124
0.0000,0.0000,0.0000,0.0000,0.0000,0.0000,0.0000
1 accuracy macro_precision macro_recall macro_f1 weighted_precision weighted_recall weighted_f1
527 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
528 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
529 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
530 0.0375 1.0000 0.0375 0.0723 1.0000 0.0375 0.0723
531 0.2281 1.0000 0.2281 0.3715 1.0000 0.2281 0.3715
532 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
533 0.0094 1.0000 0.0094 0.0186 1.0000 0.0094 0.0186
534 0.0094 1.0000 0.0094 0.0186 1.0000 0.0094 0.0186
535 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
536 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
537 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
538 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
539 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
540 0.0063 1.0000 0.0063 0.0124 1.0000 0.0063 0.0124
541 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

12
reports/WeightFiltration/RESNET34/retain.csv
View File

@@ -527,3 +527,15 @@ accuracy,macro_precision,macro_recall,macro_f1,weighted_precision,weighted_recal
0.9546,0.9558,0.9546,0.9547,0.9558,0.9546,0.9547 0.9546,0.9558,0.9546,0.9547,0.9558,0.9546,0.9547
0.9526,0.9540,0.9526,0.9528,0.9540,0.9526,0.9528 0.9526,0.9540,0.9526,0.9528,0.9540,0.9526,0.9528
0.9526,0.9540,0.9526,0.9528,0.9540,0.9526,0.9528 0.9526,0.9540,0.9526,0.9528,0.9540,0.9526,0.9528
0.9520,0.9541,0.9520,0.9524,0.9541,0.9520,0.9524
0.9520,0.9542,0.9520,0.9524,0.9542,0.9520,0.9524
0.9507,0.9515,0.9507,0.9507,0.9515,0.9507,0.9507
0.9533,0.9548,0.9533,0.9535,0.9548,0.9533,0.9535
0.9566,0.9583,0.9566,0.9569,0.9583,0.9566,0.9569
0.9572,0.9588,0.9572,0.9574,0.9588,0.9572,0.9574
0.9533,0.9551,0.9533,0.9536,0.9551,0.9533,0.9536
0.9487,0.9518,0.9487,0.9493,0.9518,0.9487,0.9493
0.9546,0.9559,0.9546,0.9547,0.9559,0.9546,0.9547
0.9520,0.9534,0.9520,0.9522,0.9534,0.9520,0.9522
0.9539,0.9554,0.9539,0.9542,0.9554,0.9539,0.9542
0.9539,0.9559,0.9539,0.9543,0.9559,0.9539,0.9543
1 accuracy macro_precision macro_recall macro_f1 weighted_precision weighted_recall weighted_f1
527 0.9546 0.9558 0.9546 0.9547 0.9558 0.9546 0.9547
528 0.9526 0.9540 0.9526 0.9528 0.9540 0.9526 0.9528
529 0.9526 0.9540 0.9526 0.9528 0.9540 0.9526 0.9528
530 0.9520 0.9541 0.9520 0.9524 0.9541 0.9520 0.9524
531 0.9520 0.9542 0.9520 0.9524 0.9542 0.9520 0.9524
532 0.9507 0.9515 0.9507 0.9507 0.9515 0.9507 0.9507
533 0.9533 0.9548 0.9533 0.9535 0.9548 0.9533 0.9535
534 0.9566 0.9583 0.9566 0.9569 0.9583 0.9566 0.9569
535 0.9572 0.9588 0.9572 0.9574 0.9588 0.9572 0.9574
536 0.9533 0.9551 0.9533 0.9536 0.9551 0.9533 0.9536
537 0.9487 0.9518 0.9487 0.9493 0.9518 0.9487 0.9493
538 0.9546 0.9559 0.9546 0.9547 0.9559 0.9546 0.9547
539 0.9520 0.9534 0.9520 0.9522 0.9534 0.9520 0.9522
540 0.9539 0.9554 0.9539 0.9542 0.9554 0.9539 0.9542
541 0.9539 0.9559 0.9539 0.9543 0.9559 0.9539 0.9543

30
reports/WeightFiltration/attack_values.csv
View File

@@ -3,3 +3,33 @@ target_class,parameter_mia_accuracy,latent_distance_tell,lookalike_accuracy
1,0.500000,1.279257,0.968750 1,0.500000,1.279257,0.968750
2,0.500000,1.717911,0.937500 2,0.500000,1.717911,0.937500
3,0.500000,1.354225,0.989583 3,0.500000,1.354225,0.989583
4,0.500000,1.308562,0.968750
5,0.500000,2.087495,0.947917
6,0.500000,1.793863,0.927083
7,0.500000,1.213634,0.937500
8,0.500000,1.705230,0.979167
9,0.500000,1.713113,0.979167
10,0.500000,1.358761,0.968750
11,0.500000,1.485312,1.000000
12,0.500000,1.393625,0.947917
13,0.500000,1.677361,0.947917
14,0.500000,1.734759,0.906250
15,0.500000,1.715959,0.968750
16,0.500000,1.180073,0.947917
17,0.500000,1.597037,0.947917
18,0.500000,1.491600,0.947917
19,0.500000,1.622301,0.968750
0,0.500000,1.562955,0.968750
0,0.500000,1.526132,0.979167
1,0.500000,1.323757,0.968750
2,0.500000,1.493928,0.958333
3,0.500000,1.643880,0.989583
4,0.500000,1.479018,0.989583
5,0.500000,2.176565,0.958333
6,0.500000,1.620670,0.958333
7,0.500000,0.977823,0.937500
8,0.500000,1.455012,0.968750
9,0.500000,2.145126,0.989583
10,0.500000,1.905214,0.895833
11,0.500000,2.171935,0.958333
12,0.500000,1.761187,0.937500
1 target_class parameter_mia_accuracy latent_distance_tell lookalike_accuracy
3 1 0.500000 1.279257 0.968750
4 2 0.500000 1.717911 0.937500
5 3 0.500000 1.354225 0.989583
6 4 0.500000 1.308562 0.968750
7 5 0.500000 2.087495 0.947917
8 6 0.500000 1.793863 0.927083
9 7 0.500000 1.213634 0.937500
10 8 0.500000 1.705230 0.979167
11 9 0.500000 1.713113 0.979167
12 10 0.500000 1.358761 0.968750
13 11 0.500000 1.485312 1.000000
14 12 0.500000 1.393625 0.947917
15 13 0.500000 1.677361 0.947917
16 14 0.500000 1.734759 0.906250
17 15 0.500000 1.715959 0.968750
18 16 0.500000 1.180073 0.947917
19 17 0.500000 1.597037 0.947917
20 18 0.500000 1.491600 0.947917
21 19 0.500000 1.622301 0.968750
22 0 0.500000 1.562955 0.968750
23 0 0.500000 1.526132 0.979167
24 1 0.500000 1.323757 0.968750
25 2 0.500000 1.493928 0.958333
26 3 0.500000 1.643880 0.989583
27 4 0.500000 1.479018 0.989583
28 5 0.500000 2.176565 0.958333
29 6 0.500000 1.620670 0.958333
30 7 0.500000 0.977823 0.937500
31 8 0.500000 1.455012 0.968750
32 9 0.500000 2.145126 0.989583
33 10 0.500000 1.905214 0.895833
34 11 0.500000 2.171935 0.958333
35 12 0.500000 1.761187 0.937500

1
reports/base/RESNET34/Finetuned.csv
View File

@@ -52,3 +52,4 @@ accuracy,macro_precision,macro_recall,macro_f1,weighted_precision,weighted_recal
0.9581,0.9590,0.9581,0.9581,0.9590,0.9581,0.9581 0.9581,0.9590,0.9581,0.9581,0.9590,0.9581,0.9581
0.9513,0.9525,0.9512,0.9514,0.9525,0.9513,0.9514 0.9513,0.9525,0.9512,0.9514,0.9525,0.9513,0.9514
0.9531,0.9541,0.9531,0.9532,0.9541,0.9531,0.9532 0.9531,0.9541,0.9531,0.9532,0.9541,0.9531,0.9532
0.9525,0.9542,0.9525,0.9528,0.9542,0.9525,0.9528
1 accuracy macro_precision macro_recall macro_f1 weighted_precision weighted_recall weighted_f1
52 0.9581 0.9590 0.9581 0.9581 0.9590 0.9581 0.9581
53 0.9513 0.9525 0.9512 0.9514 0.9525 0.9513 0.9514
54 0.9531 0.9541 0.9531 0.9532 0.9541 0.9531 0.9532
55 0.9525 0.9542 0.9525 0.9528 0.9542 0.9525 0.9528

2
unlearning/WeightFiltration.py
View File

@@ -126,7 +126,7 @@ class WeightFiltration(Strategy):
print(f"Gating matrix loaded. Switching layout to target class index: {self.target_class_index}") print(f"Gating matrix loaded. Switching layout to target class index: {self.target_class_index}")
self.wf_model.target_class_index = self.target_class_index self.wf_model.target_class_index = self.target_class_index
return self.wf_model.get() return self.wf_model
def _split_data(self, dataset): def _split_data(self, dataset):
return vertical_split( return vertical_split(