ryderling DEEPSEC issues - Githubissues

ryderling / DEEPSEC

DEEPSEC: A Uniform Platform for Security Analysis of Deep Learning Model

MIT License

206 stars 71 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

What's the difference between UMIFGSM and TMIFGSM

#16 XiaodanLi001 opened 4 years ago
2
PGD/BIM implementation is incorrect

#15 carlini opened 5 years ago
1
JSMA implementation is incorrect

#14 carlini opened 5 years ago
0
Significant and fundamental flaws in methodology, analysis, and conclusions

#13 carlini opened 5 years ago
1
Discrepancies between tables, text, and code

#12 carlini opened 5 years ago
7
Paper does not report attack success rate for targeted adversarial examples

#11 carlini opened 5 years ago
2
Reporting success rate of unbounded attacks is meaningless

#10 carlini opened 5 years ago
2
Attack success rate decreases with distortion bound

#9 carlini opened 5 years ago
2
Detection defenses set per-attack thresholds

#8 carlini closed 5 years ago
2
Epsilon values studied are too large to be meaningful

#7 carlini opened 5 years ago
4
Comparing attack effectiveness is done incorrectly

#6 carlini opened 5 years ago
2
Computing the average over different threat models is meaningless

#5 carlini opened 5 years ago
2
PGD adversarial training implementation is incorrect

#4 carlini opened 5 years ago
3
FGSM implementation is incorrect

#3 carlini opened 5 years ago
9
Paper uses averages instead of the minimum for security analysis

#2 carlini opened 5 years ago
2
Attacks are not run on defenses in an all-pairs manner

#1 carlini opened 5 years ago
2