subpopulation poisoning attacks

We introduce a method to manipulate neuron activations while pre-training models, allowing highly successful inference of sensitive properties of the victim’s downstream training data.

We introduce a method to manipulate neuron activations while pre-training models, allowing highly successful inference of sensitive properties of the victim’s downstream training data.

We propose efficient data poisoning attacks that can asymptotically approach a target model with desired properties.

We propose efficient data poisoning attacks that can asymptotically approach a target model with desired properties.