DeepSecure: Scalable Provably-Secure Deep Learning

By Bita Darvish Rouhani, M. Sadegh Riazi, Farinaz Koushanfar. 2017

452 citations recorded by the source corpus.

This paper presents DeepSecure, the an scalable and provably secure Deep Learning (DL) framework that is built upon automated design, efficient logic synthesis, and optimization methodologies. DeepSecure targets scenarios in which neither of the involved parties including the cloud servers that hold the DL model parameters or the delegating clients who own the data is willing to reveal their information. Our framework is the first to empower accurate and scalable DL analysis of data generated by distributed clients without sacrificing the security to maintain efficiency. The secure DL computation in DeepSecure is performed using Yao's Garbled Circuit (GC) protocol. We devise GC-optimized realization of various components used in DL. Our optimized implementation achieves up to 58-fold higher throughput per sample compared with the best prior solution. In addition to the optimized GC realization, we introduce a set of novel low-overhead pre-processing techniques which further reduce the GC overall runtime in the context of DL. Our extensive evaluations demonstrate up to two orders-of-magnitude additional runtime improvement achieved as a result of our pre-processing methodology.

Subfields: Artificial Intelligence, Electrical and Electronic Engineering, Safety, Robustness & Interpretability, Privacy & Security

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Electrical and Electronic Engineering

In Safety, Robustness & Interpretability

In Privacy & Security

Citations in the corpus

Federated Machine Learning (2019, background): “e the privacy of gradients and enhance the security of the system. With the recent advances in deep learning, privacy-preserving neural networks inference is also receiving a lot of research interests[10,11,14,28,40,52,54]. ACM Trans. Intell. Syst. Technol., Vol. 10, No. 2, Article 12. Publication d”
Federated Learning: Challenges, Methods, and Future Directions (2019, background): “mmunication and computation costs. Moreover, SMC protocols need to be carefully designed and implemented for each operation in the targeted learning algorithm [25, 79]. We defer interested readers to [13, 97] for a more comprehensive review of the approaches based on homomorphic encryption and SMC. ”
Style transfer-based image synthesis as an efficient regularization technique in deep learning (2019, background): “We paid special attention to avoid a situation where any images that were used to generate the augmented database did not appear in the validation or test sets (data leakage phenomena [39]).”
CryptoDL: Deep Neural Networks over Encrypted Data (2017, methodology): “[50] and [39] are two recent works based on SMC techniques and were discussed in section 4.1.”
Gazelle: A Low Latency Framework for Secure Neural Network Inference (2018, methodology): “3) C : 1-Conv and 2-FC layers with ReLU activation from [33].”
Chameleon: A Hybrid Secure Computation Framework for Machine Learning Applications (2018, methodology): “The DeepSecure framework [75] is a GC-based framework for secure Deep Learning inference.”
Privacy-preserving Machine Learning as a Service (2018, neutral): “For classifying a batch input with size 8192, the running times for [41] and [25] are 10649 and 10485.”
CrypTFlow2: Practical 2-Party Secure Inference (2020, methodology): “Let 1 { 𝑏 } denote the indicator function that is 1 when 𝑏 is true and 0 when 𝑏 is false .”
EPIC: Efficient Private Image Classification (or: Learning from the Masters) (2019, methodology): “DeepSecure [45] is another work in the 2-PC setting, and the passive security model, using Garbled-Circuit techniques.”
A Hybrid Deep Learning Architecture for Privacy-Preserving Mobile Analytics (2017, background): “[46] and Mohassel and Zhang [47] tried to”
Secure Byzantine-Robust Machine Learning (2020, background): “[14, 33, 15, 22, 26, 33, 10, 16, 32] falls into a general category of 2-party computation (2PC).”
ML-Leaks: Model and Data Independent Membership Inference Attacks and Defenses on Machine Learning Models (2018, background): “Besides the above, many recent works have tackled security and privacy in machine learning from various perspectives [39], [16], [9], [37], [7], [27], [20], [45], [36], [21], [40], [29], [41], [20].”
Security and Privacy Issues in Deep Learning (2018, methodology): “2017], DeepSecure [Rouhani et al. 2018] and Gazelle [Juvekar et al.”