提出高效加密推理框架，实现高吞吐、低延迟、省内存的隐私保护神经网络计算
Towards Deep Encrypted Training: Low-Latency, Memory-Efficient, and High-Throughput Inference for Privacy-Preserving Neural Networks
- 设计批处理优化算法与资源高效流水线架构
- 512批次下每图推理仅8.86秒，内存降低3.74倍
- 适合需要批量加密推理的医疗/金融等敏感场景
隐私保护机器学习在需保密敏感数据的应用中日益重要。同态加密（HE）可直接对加密数据进行计算，实现无需暴露原始输入的神经网络推理。现有工作多集中于单张加密图像的推理，而批处理加密输入的性能仍滞后，尽管这对高吞吐推理和训练相关任务至关重要。本文通过优化批处理的HE友好型神经网络算法，填补这一空白。我们还提出一种针对不同批大小的资源高效流水线架构。在加密的CIFAR-10和CIFAR-100数据集上，分别使用HE友好型ResNet-20和ResNet-34模型进行了实现与评估。对于ResNet-20，在512个加密图像的批处理下，平均推理时间为每图8.86秒，峰值内存占用98.96GB，相比当前最优方案，运行时间提升1.78倍，内存减少3.74倍。对于更深的ResNet-34模型，在256个加密图像批处理下，平均推理时间为每图28.14秒，使用246.78GB内存。
原文摘要 · Abstract (English)
Privacy-preserving machine learning (PPML) has become increasingly important in applications where sensitive data must remain confidential. Homomorphic Encryption (HE) enables computation directly on encrypted data, allowing neural network inference without revealing raw inputs. While prior works have largely focused on inference over a single encrypted image, batch processing of encrypted inputs lags behind, despite being critical for high-throughput inference scenarios and training-oriented workloads. In this work, we address this gap by developing optimized algorithms for batched HE-friendly neural networks. We also introduced a pipeline architecture designed to maximize resource efficiency for different batch size execution. We implemented these algorithms and evaluated our work using HE-friendly ResNet-20 and ResNet-34 models on encrypted CIFAR-10 and CIFAR-100 datasets, respectively. For ResNet-20, our approach achieves an amortized inference time of 8.86 seconds per image when processing a batch of 512 encrypted images, with a peak memory usage of 98.96 GB. These results represent a 1.78x runtime improvement and a 3.74x reduction in memory usage compared to the state-of-the-art design. For the deeper ResNet-34 model, we achieve an amortized inference time of 28.14 on a batch of 256 encrypted images using 246.78GB of RAM
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。