Khawaja Abaid Ullah

ADS-C: Antidistillation Sampling for Classification

Khawaja Abaid Ullah, Mohammad Javad Khojasteh

Knowledge distillation enables an adversary to replicate a proprietary classifier by querying its prediction interface and training a surrogate on the returned probability vectors. Antidistillation sampling, proposed for large language models, counters this threat with an input-d…Read more