Crypto-Oriented Neural Architecture Design
This addresses privacy conflicts in neural network applications by making secure inference more efficient, though it is incremental as it builds on existing architectures.
The paper tackles the inefficiency of secure computation for neural networks by proposing a crypto-oriented neural architecture with a novel Partial Activation layer, which significantly improves the speed of secure inference on three state-of-the-art architectures.
As neural networks revolutionize many applications, significant privacy conflicts between model users and providers emerge. The cryptography community developed a variety of techniques for secure computation to address such privacy issues. As generic techniques for secure computation are typically prohibitively ineffective, many efforts focus on optimizing their underlying cryptographic tools. Differently, we propose to optimize the initial design of crypto-oriented neural architectures and provide a novel Partial Activation layer. The proposed layer is much faster for secure computation. Evaluating our method on three state-of-the-art architectures (SqueezeNet, ShuffleNetV2, and MobileNetV2) demonstrates significant improvement to the efficiency of secure inference on common evaluation metrics.