Timezone: »

 
Poster
Rethinking Backdoor Attacks
Alaa Khaddaj · Guillaume Leclerc · Aleksandar Makelov · Kristian Georgiev · Hadi Salman · Andrew Ilyas · Aleksander Madry

Tue Jul 25 02:00 PM -- 04:30 PM (PDT) @ Exhibit Hall 1 #710
in Poster Session 1 »

In a backdoor attack, an adversary inserts maliciously constructed backdoor examples into a training set to make the resulting model vulnerable to manipulation. Defending against such attacks involves viewing inserted examples as outliers in the training set and using techniques from robust statistics to detect and remove them. In this work, we present a different approach to the backdoor attack problem. Specifically, we show that without structural information about the training data distribution, backdoor attacks are indistinguishable from naturally-occuring features in the data---and thus impossible to "detect" in a general sense. Then, guided by this observation, we revisit existing defenses against backdoor attacks and characterize the (often latent) assumptions they make, and on which they depend. Finally, we explore an alternative perspective on backdoor attacks: one that assumes these attacks correspond to the strongest feature in the training data. Under this assumption (which we make formal) we develop a new primitive for detecting backdoor attacks. Our primitive naturally gives rise to a detection algorithm that comes with theoretical guarantees, and is effective in practice.

[ PDF] 

Author Information

Alaa Khaddaj (MIT)
Guillaume Leclerc (MIT)
Aleksandar Makelov (Massachusetts Institute of Technology)
Kristian Georgiev (MIT)
Hadi Salman (OpenAI / MIT)
Andrew Ilyas (MIT)
Aleksander Madry (MIT)

More from the Same Authors