Skip to content

Instantly share code, notes, and snippets.

@seanwatson

seanwatson/param_comparison.csv

Created November 1, 2017 00:06
Show Gist options
  • Save seanwatson/65855b634a5601cbe504dd61c0d8154b to your computer and use it in GitHub Desktop.
Save seanwatson/65855b634a5601cbe504dd61c0d8154b to your computer and use it in GitHub Desktop.
Download ZIP
Comparison of FGSM Adversarial Training Across MNIST, CIFAR-10, and CIFAR-100 with Different Model Parameters
Raw
param_comparison.csv
epoch filters type clean_train_clean_eval clean_train_adv_eval adv_train_clean_eval adv_train_adv_eval
2 64 mnist 0.9897 0.0897 0.9871 0.9002
2 64 cifar_10 0.4564 0.079 0.4166 0.1692
2 64 cifar_100 0.2613 0.0124 0.2399 0.0642
3 64 mnist 0.9926 0.0688 0.9893 0.934
3 64 cifar_10 0.4476 0.0961 0.4472 0.3031
3 64 cifar_100 0.2777 0.0159 0.2668 0.0791
4 64 mnist 0.9921 0.089 0.9918 0.9362
4 64 cifar_10 0.5018 0.0821 0.4577 0.3609
4 64 cifar_100 0.3207 0.0191 0.293 0.1245
5 64 mnist 0.9932 0.1083 0.9908 0.9187
5 64 cifar_10 0.5085 0.0917 0.4971 0.4082
5 64 cifar_100 0.3063 0.0172 0.3084 0.1139
6 64 mnist 0.9933 0.1219 0.9913 0.916
6 64 cifar_10 0.517 0.1062 0.4914 0.4363
6 64 cifar_100 0.338 0.02 0.3205 0.1486
7 64 mnist 0.993 0.0753 0.9913 0.9554
7 64 cifar_10 0.5238 0.1152 0.5017 0.458
7 64 cifar_100 0.342 0.0196 0.3271 0.1568
8 64 mnist 0.9931 0.0899 0.9909 0.9537
8 64 cifar_10 0.5264 0.1203 0.5195 0.4698
8 64 cifar_100 0.3369 0.0171 0.3221 0.1495
9 64 mnist 0.9934 0.1252 0.9913 0.9356
9 64 cifar_10 0.5284 0.1366 0.4923 0.4305
9 64 cifar_100 0.3369 0.0237 0.3355 0.1743
10 64 mnist 0.9929 0.1568 0.9925 0.9517
10 64 cifar_10 0.5437 0.1262 0.5176 0.4777
10 64 cifar_100 0.341 0.0201 0.3382 0.1801
6 16 mnist 0.9912 0.0294 0.9825 0.8808
6 16 cifar_10 0.4736 0.0762 0.4547 0.3475
6 16 cifar_100 0.2625 0.0162 0.2337 0.0792
6 32 mnist 0.9922 0.0287 0.9892 0.909
6 32 cifar_10 0.5051 0.1174 0.4883 0.3899
6 32 cifar_100 0.3073 0.0189 0.2855 0.1178
6 64 mnist 0.9923 0.1171 0.9902 0.9194
6 64 cifar_10 0.5219 0.121 0.4846 0.3953
6 64 cifar_100 0.334 0.0194 0.3416 0.1759
6 128 mnist 0.9925 0.2074 0.9928 0.9677
6 128 cifar_10 0.5048 0.1193 0.4842 0.43
6 128 cifar_100 0.3543 0.0231 0.3448 0.1668
6 256 mnist 0.992 0.1527 0.9923 0.9695
6 256 cifar_10 0.5358 0.0937 0.5104 0.4783
6 256 cifar_100 0.3502 0.019 0.3544 0.2059
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment