Securely Outsourcing Neural Network Inference to the Cloud With Lightweight Techniques

Liu, Xiaoning; Zheng, Yifeng; Yuan, Xingliang; Yi, Xun

doi:10.1109/tdsc.2022.3141391

Cited by 26 publications

(11 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Besides, we mainly measure the performance of the SecCom operations. As shown in Table II, we see that the overheads of communication and the runtime of inference for the secure comparison (SecCom) protocol are notably less than the prior works [22], [33], [34]. We analyze that our secure comparison protocol over the 32-bit number achieves 2×, 2×, 75.6× bandwidth savings and 3.5×, 2×, 30.6× runtime savings compared with the prior works [22], [33], [34].…”

Section: Microbenchmarksmentioning

confidence: 85%

“…As shown in Table II, we see that the overheads of communication and the runtime of inference for the secure comparison (SecCom) protocol are notably less than the prior works [22], [33], [34]. We analyze that our secure comparison protocol over the 32-bit number achieves 2×, 2×, 75.6× bandwidth savings and 3.5×, 2×, 30.6× runtime savings compared with the prior works [22], [33], [34]. To ensure our secure protocols have strong scalability, we put n numbers into an array of Pytorch to execute vectorization operations in parallel instead of doing secure comparisons one by one.…”

Section: Microbenchmarksmentioning

confidence: 87%

“…Similar to prior works [22], [23] in the two-cloud-server setting, we assume that the model user, the model owner, and the two independent cloud servers from the different cloud providers all are semi-honest and non-colluding. In other words, each server will allow all protocols exactly as specified yet may attempt to learn the privacy information of the user or model owner from their known data.…”

Section: B Threat Modelmentioning

confidence: 99%

“…SCUT-HEAD is a large-scale head detection dataset, including 4405 images labeld with 111251 heads. Besides, we evaluate L-SecNet on several neural network architectures including 3FC-ReLU-FC model architecture (Model 1) specified in Sonic [22] , AlexNet(Model 2), Logistic Regression specified in SecureML [9], FitNet(Model-3) specified in Sonic [22], ResNet-32 [35] (Model-4) specified in DELPHI [10], ResNet-50 (Model-5) specified in CrypTFlow [10], Heads model (Model-6) with CNNs and RNNs specified in SIRNN [26].…”

Section: A Experiments Setupmentioning

confidence: 99%

See 3 more Smart Citations

L-SecNet: Towards Secure and Lightweight Deep Neural Network Inference

Song,

Fu,

et al. 2023

J-NaNA

View full text Add to dashboard Cite

The advances in machine learning technology has promoted its great potential for deep neural network (DNN) inference powered applications of Internet of Things (IoT), such as facial verification cameras and speech recognition assistants. The current deployment of these applications also raises serious privacy concerns, especially when sensitive individual information is accessed easily by various IoT devices. Fortunately, the cryptography-based solutions are able to execute secure inference without infringing the user’s raw data and the model owner’s proprietary model. However, existing works suffer from impractically high latency and low accuracy, stemming primarily from the evaluations of the non-linear layers in DNN. In this paper, we propose L-SecNet, a lightweight secure neural network inference system that provides efficient inference services without sacrificing accuracy and privacy. Specifically, to reduce latency caused by comparison operations in non-linear layers, we subtly combine additive secret sharing and multiplicative secret sharing to design a lightweight secure comparison protocol. Further, we approximate the commonly used and time-consuming activation functions (including Sigmoid and Tanh functions) with the non-linear sin function instead of the linear polynomial approximation functions in the existing works. In order to maintain low running latency while meeting the requirement of high accuracy, a secure and lightweight protocol for a sin function is proposed. Our theoretical analysis and empirical experiments evaluate the security and efficiency of the L-SecNet system. Compared with the state-of-the-art works, L-SecNet saves up to about 80 times bandwidth and about 53 times runtime.

show abstract

Section: Microbenchmarksmentioning

confidence: 85%

Section: Microbenchmarksmentioning

confidence: 87%

Section: B Threat Modelmentioning

confidence: 99%

Section: A Experiments Setupmentioning

confidence: 99%

See 2 more Smart Citations

L-SecNet: Towards Secure and Lightweight Deep Neural Network Inference

Song,

Fu,

et al. 2023

J-NaNA

View full text Add to dashboard Cite

show abstract

“…Compared to previous schemes based on additive secret-sharing techniques, Falcon utilizes three-party replicated secret-sharing techniques to achieve faster runtimes and lower communication overhead. Sonic [20] designs a series of secure and efficient neural network layer functions based on additive secret sharing [18]. Sonic's secure computation protocol runs between two computational parties.…”

Section: Related Workmentioning

confidence: 99%

PPTIF: Privacy-Preserving Transformer Inference Framework for Language Translation

Liu,

2024

IEEE Access

View full text Add to dashboard Cite

The Transformer model has emerged as a prominent machine learning tool within the field of natural language processing. Nevertheless, running the Transformer model on resource-constrained devices presents a notable challenge that needs to be addressed. Although outsourcing services can significantly reduce the computational overhead associated with using the model, it also incurs privacy risks to the provider's proprietary model and the client's sensitive data. In this paper, we propose an efficient privacypreserving Transformer inference framework (PPTIF) for language translation tasks based on three-party replicated secret-sharing techniques. PPTIF offers a secure approach for users to leverage Transformer-based applications, such as language translation, while maintaining the confidentiality of their original input and inference results, thereby preventing any disclosure to the cloud server. Meanwhile, PPTIF ensures robust protection for the model parameters, guaranteeing their integrity and confidentiality. In PPTIF, we design a series of interaction protocols to implement the secure computation of Transformer components, namely secure Encoder and secure Decoder. To improve the efficiency of PPTIF, we optimize the computation of the Scaled Dot-Product Attention (Transformer's core operation) under secret sharing, effectively reducing its computation and communication overhead. Compared with Privformer, the optimized Masked Multi-Head Attention achieves about 1.7× lower runtime and 2.3× lower communication. In total, PPTIF achieve about 1.3× lower runtime and 1.2× lower communication. The effectiveness and security of PPTIF have been rigorously evaluated through comprehensive theoretical analysis and experimental validation.

show abstract

A reputation-based and privacy-preserving incentive scheme for mobile crowd sensing: a deep reinforcement learning approach

Zhang

Shi

et al. 2022

Wireless Netw

View full text Add to dashboard Cite

Securely Outsourcing Neural Network Inference to the Cloud With Lightweight Techniques

Cited by 26 publications

References 40 publications

L-SecNet: Towards Secure and Lightweight Deep Neural Network Inference

L-SecNet: Towards Secure and Lightweight Deep Neural Network Inference

PPTIF: Privacy-Preserving Transformer Inference Framework for Language Translation

A reputation-based and privacy-preserving incentive scheme for mobile crowd sensing: a deep reinforcement learning approach

Contact Info

Product

Resources

About