Primer: Fast Private Transformer Inference on Encrypted Data. (arXiv:2303.13679v1 [cs.CR])

It is increasingly important to enable privacy-preserving inference for cloud
services based on Transformers. Post-quantum cryptographic techniques, e.g.,
fully homomorphic encryption (FHE), and multi-party computation (MPC), are
popular methods to support private Transformer inference. However, existing
works still suffer from prohibitively computational and communicational
overhead. In this work, we present, Primer, to enable a fast and accurate
Transformer over encrypted data for natural language processing tasks. In
particular, Primer is constructed by a hybrid cryptographic protocol optimized
for attention-based Transformer …

attention cloud cloud services computation computational data enable encrypted encrypted data encryption fast fhe fully homomorphic encryption homomorphic encryption hybrid important language merge mpc natural language natural language processing party popular post-quantum privacy private protocol quantum services support techniques tokens transformers work