Video
Unknown · 0:00
Skyler (USC) presents the N-gram Coverage Attack, a black-box membership inference attack (MIA). It tests whether a language model was trained on a given document using only the model's text outputs. The model is give...
Unknown · 0:00
Skyler (USC) presents the N-gram Coverage Attack, a black-box membership inference attack (MIA). It tests whether a language model was trained on a given document using only the model's text outputs. The model is give...
Redirecting...