NASS: Optimizing Secure Inference via Neural Architecture Search
This addresses privacy concerns in secure inference for clients and servers by enhancing performance, though it is incremental as it builds on existing secure protocols.
The paper tackled the problem of optimizing neural network architectures for secure inference to balance accuracy and efficiency, achieving an improvement in prediction accuracy from 81.6% to 84.6% while reducing inference runtime by 2x and communication bandwidth by 1.9x on CIFAR-10.
Due to increasing privacy concerns, neural network (NN) based secure inference (SI) schemes that simultaneously hide the client inputs and server models attract major research interests. While existing works focused on developing secure protocols for NN-based SI, in this work, we take a different approach. We propose NASS, an integrated framework to search for tailored NN architectures designed specifically for SI. In particular, we propose to model cryptographic protocols as design elements with associated reward functions. The characterized models are then adopted in a joint optimization with predicted hyperparameters in identifying the best NN architectures that balance prediction accuracy and execution efficiency. In the experiment, it is demonstrated that we can achieve the best of both worlds by using NASS, where the prediction accuracy can be improved from 81.6% to 84.6%, while the inference runtime is reduced by 2x and communication bandwidth by 1.9x on the CIFAR-10 dataset.