Training Set Camouflage
share
We introduce a form of steganography in the domain of machine learning which we call training set camouflage. Imagine Alice has a training set on an illicit machine learning classification task. Alice wants Bob (a machine learning system) to learn the task. However, sending either the training set or the trained model to Bob can raise suspicion if the communication is monitored. Training set camouflage allows Alice to compute a second training set on a completely different -- and seemingly benign -- classification task. By construction, sending the second training set will not raise suspicion. When Bob applies his standard (public) learning algorithm to the second training set, he approximately recovers the classifier on the original task. Training set camouflage is a novel form of steganography in machine learning. We formulate training set camouflage as a combinatorial bilevel optimization problem and propose solvers based on nonlinear programming and local search. Experiments on real classification tasks demonstrate the feasibility of such camouflage.
READ FULL TEXT