Download presentation

Presentation is loading. Please wait.

Published byNico Potter Modified over 3 years ago

1
Spam Deobfuscation Wissam Kazan Daniel Woods

2
Problem Description Example Spam Text: Get your Viagra pills here When Obfuscated: G3t y0u*r \/|aGrra Pi11z |-|eer Problem: Reverse Obfuscation so Spam Filters Work

3
Decoding G3t y0u*r \/|aGrra Pi11z |-|eer xxxxxxxxxxxxxxx?xxxxxxxxxxxxxxx Alignment G3t y0u*r \/|aGrra Pi11z |-|eer Get you-r V-iagr-a pills h--ere Solution Breakdown

4
Alignment Example Original Text: Wissam Obfuscated Text: VV|sS/\m Naive Alignment: Wissa--m EM Alignment: W-issa-m Correct Alignment: W-issa-m (Result 1.4% error at 37.7% obfuscation)

5
Example: G3t y0u*r \/|aGrra Pi11z |-|eer Maximum Likelihood HMM (Viterbi) G -> e -> t -> _ -> y –> o ->... | | | G 3 t _ y 0... Decoding Approaches Get y xx?xx Get y xx?xx Get_y xx?xx Get y xx?xx

6
Results (37.7% Obfuscation)

7
Example Result Obfuscated Text: a"bbove diGs{lqaiim3rss and+ e+XclAusXi70Nsd may nQoSt Apxply Correct Test: above disclaimers and exclusions may not apply HMM Prediction: abbove dislaimerss and tclusiond may not pply

Similar presentations

OK

HMM II: Parameter Estimation. Reminder: Hidden Markov Model Markov Chain transition probabilities: p(S i+1 = t|S i = s) = a st Emission probabilities:

HMM II: Parameter Estimation. Reminder: Hidden Markov Model Markov Chain transition probabilities: p(S i+1 = t|S i = s) = a st Emission probabilities:

© 2018 SlidePlayer.com Inc.

All rights reserved.

To make this website work, we log user data and share it with processors. To use this website, you must agree to our Privacy Policy, including cookie policy.

Ads by Google