CrypTFlow: Secure TensorFlow Inference

@article{Kumar2020CrypTFlowST,
  title={CrypTFlow: Secure TensorFlow Inference},
  author={Nishant Kumar and Mayank Rathee and Nishanth Chandran and Divya Gupta and Aseem Rastogi and Rahul Sharma},
  journal={2020 IEEE Symposium on Security and Privacy (SP)},
  year={2020},
  pages={336-353}
}
We present CrypTFlow, a first of its kind system that converts TensorFlow inference code into Secure Multi-party Computation (MPC) protocols at the push of a button. To do this, we build three components. Our first component, Athos, is an end-to-end compiler from TensorFlow to a variety of semihonest MPC protocols. The second component, Porthos, is an improved semi-honest 3-party protocol that provides significant speedups for TensorFlow like applications. Finally, to provide malicious secure… Expand
CrypTFlow2: Practical 2-Party Secure Inference
TLDR
Using CrypTFlow2, the first secure inference over ImageNet-scale DNNs like ResNet50 and DenseNet121 is presented, at least an order of magnitude larger than those considered in the prior work of 2-party DNN inference. Expand
Secure Medical Image Analysis with CrypTFlow
TLDR
This paper provides the first evaluation of secure segmentation of 3D images, a task that requires much more powerful models than classification and is the largest secure inference task run till date. Expand
CrypTen: Secure Multi-Party Computation Meets Machine Learning
Secure multi-party computation (MPC) allows parties to perform computations on data while keeping that data private. This capability has great potential for machine-learning applications: itExpand
Fantastic Four: Honest-Majority Four-Party Secure Computation With Malicious Security
TLDR
The results show that the four- party protocol performs similarly to an efficient honest-majority three-party protocol that only provides semi-honest/passive security, which suggests that adding a fourth party can be an effective method to achieve active security without harming performance. Expand
secureTF: A Secure TensorFlow Framework
Data-driven intelligent applications in modern online services have become ubiquitous. These applications are usually hosted in the untrusted cloud computing infrastructure. This poses significantExpand
MUSE: Secure Inference Resilient to Malicious Clients
The increasing adoption of machine learning inference in applications has led to a corresponding increase in concerns surrounding the privacy guarantees offered by existing mechanisms for inference.Expand
Adam in Private: Secure and Fast Training of Deep Neural Networks with Adaptive Moment Estimation
TLDR
This work proposes a framework that allows efficient and secure evaluation of full-fledged state-of-the-art ML algorithms via secure multiparty computation (MPC), and proposes secure and efficient protocols for the above seemingly MPC-unfriendly computations (but which are essential to DNN). Expand
MP-SPDZ: A Versatile Framework for Multi-Party Computation
TLDR
The variety of protocols implemented and the design choices made in the development of MP-SPDZ are outlined as well as the capabilities of the programming interface. Expand
Accelerating 2PC-based ML with Limited Trusted Hardware
TLDR
Otak improves efficiency via a new 2PC protocol that tailors recent primitives such as function and homomorphic secret sharing to ML inference, and uses trusted hardware in a limited capacity to bootstrap the protocol. Expand
Secure Evaluation of Quantized Neural Networks
TLDR
This work shows that the separating line between existing ML frameworks and existing MPC protocols may be narrower than implicitly suggested by previous works. Expand
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 103 REFERENCES
Global-Scale Secure Multiparty Computation
TLDR
This work designs an efficient preprocessing phase that allows the parties to generate authenticated information; it shows how to use this information to distributively construct a single "authenticated" garbled circuit that is evaluated by one party. Expand
LevioSA: Lightweight Secure Arithmetic Computation
TLDR
This work design, optimize, and implement anactively secure protocol for secure two-party arithmetic computation, which can make a fully modular black-box use of any passively secure implementation of oblivious linear function evaluation (OLE). Expand
SecureNN: 3-Party Secure Computation for Neural Network Training
TLDR
This work provides novel three-party secure computation protocols for various NN building blocks such as matrix multiplication, convolutions, Rectified Linear Units, Maxpool, normalization and so on, which is the first system to provide any security against malicious adversaries for the secure computation of complex algorithms such as neural network inference and training. Expand
High-Throughput Semi-Honest Secure Three-Party Computation with an Honest Majority
TLDR
A new information-theoretic protocol for secure three-party computation with an honest majority and a Kerberos extension that replaces the ticket-granting-ticket encryption on the Key Distribution Center in MIT-Kerberos with the protocol, using keys/ passwords that are shared between the servers. Expand
Private Machine Learning in TensorFlow using Secure Computation
TLDR
This work presents a framework for experimenting with secure multi-party computation directly in TensorFlow, gives an open source implementation of a state-of-the-art protocol and reports on concrete benchmarks using typical models from private machine learning. Expand
Optimizing Authenticated Garbling for Faster Secure Two-Party Computation
TLDR
Several optimizations of their protocol for generating AND triples result in a significant improvement in the overall communication and running time and make the communication of the online phase essentially equivalent to that of state-of-the-art semi-honest secure computation. Expand
CHET: an optimizing compiler for fully-homomorphic neural-network inferencing
TLDR
CHET is a domain-specific optimizing compiler designed to make the task of programming FHE applications easier, and generates homomorphic circuits that outperform expert-tuned circuits and makes it easy to switch across different encryption schemes. Expand
MP-SPDZ: A Versatile Framework for Multi-Party Computation
TLDR
The variety of protocols implemented and the design choices made in the development of MP-SPDZ are outlined as well as the capabilities of the programming interface. Expand
Secure Evaluation of Quantized Neural Networks
TLDR
This work shows that the separating line between existing ML frameworks and existing MPC protocols may be narrower than implicitly suggested by previous works. Expand
Outsourcing Private Machine Learning via Lightweight Secure Arithmetic Computation
TLDR
This work proposes an actively secure protocol for outsourcing secure and private machine learning computations and showcases the efficiency of the protocol by applying it to real-world instances of arithmetized neural network computations, including a network trained to perform collaborative disease prediction. Expand
...
1
2
3
4
5
...