Training-Time Attacks

What Distributions are Robust to Indiscriminate Poisoning Attacks for Linear Learners?

What Distributions are Robust to Indiscriminate Poisoning Attacks for Linear Learners?

Understand the inherent vulnerabilities to indiscriminate data poisoning attacks for linear learners by studying the optimal poisoning strategy from the perspective of data distribution.

Fnu Suya, Xiao Zhang, Yuan Tian, David Evans