We initiate the study of private inference on Transformer-based models in the client-server
setting, where clients have private inputs and servers hold proprietary models. Our main …

Abstract The growth of Graph Convolution Network (GCN) model sizes has revolutionized
numerous applications, surpassing human performance in areas such as personal …

The growth of the Machine-Learning-As-A-Service (MLaaS) market has highlighted clients'
data privacy and security issues. Private inference (PI) techniques using cryptographic …

We studied 53 privacy-preserving neural-network papers in 2016-2022 based on
cryptography (without trusted processors or differential privacy), 16 of which only use …

As the application of deep learning continues to grow, so does the amount of data used to
make predictions. While traditionally big-data deep learning was constrained by computing …

The recent rise of privacy concerns has led researchers to devise methods for private neural
inference—where inferences are made directly on encrypted data, never seeing inputs. The …

Private inference (PI) has emerged as a promising solution to execute computations on
encrypted data, safeguarding user privacy and model parameters in edge computing …

Secure transformer inference has emerged as a prominent research topic following the
proliferation of ChatGPT. Existing solutions are typically interactive, involving substantial …

Private inference (PI) enables inferences directly on cryptographically secure data. While
promising to address many privacy issues, it has seen limited use due to extreme runtimes …

The large number of ReLU non-linearity operations in existing deep neural networks makes
them ill-suited for latency-efficient private inference (PI). Existing techniques to reduce ReLU …