Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.
AICS 17
DR. SVEN KRASSER CHIEF SCIENTIST
@SVENKRASSER
SELECTED R&D
§ Static Analysis
§ Malware classification based on file traits
§ http://tinyurl.com/aics-static
§ Dynami...
ALMOST PERFECT IS
NOT GOOD ENOUGH
0.01% FALSE POSITIVES
§ Assume 100,000 clean executables
on a system
§ Single malware file
§ 99% TPR @ 0.01% FPR
§ ~1 ...
99% TRUE POSITIVES
5
Chanceofatleastone
successforadversary
Number of attempts
1%
>99%
500
AICS 2017
Upcoming SlideShare
Loading in …5
×

AICS 2017

233 views

Published on

Slides from the “Applications of AI to Operations” panel at the AAAI-17 Workshop on Artificial Intelligence for Cyber Security (AICS).

Workshop program: http://www-personal.umich.edu/~arunesh/AICS2017/program.html

Published in: Data & Analytics
  • Be the first to comment

  • Be the first to like this

AICS 2017

  1. 1. AICS 17 DR. SVEN KRASSER CHIEF SCIENTIST @SVENKRASSER
  2. 2. SELECTED R&D § Static Analysis § Malware classification based on file traits § http://tinyurl.com/aics-static § Dynamic Analysis § Malware classification based on large scale event data § http://tinyurl.com/aics-dynamic
  3. 3. ALMOST PERFECT IS NOT GOOD ENOUGH
  4. 4. 0.01% FALSE POSITIVES § Assume 100,000 clean executables on a system § Single malware file § 99% TPR @ 0.01% FPR § ~1 TPR § 10 FPRs § 91% of detections are false
  5. 5. 99% TRUE POSITIVES 5 Chanceofatleastone successforadversary Number of attempts 1% >99% 500

×