Chameleon: A Hybrid Secure Computation Framework for Machine Learning Applications

By M. Sadegh Riazi, Christian Weinert, Oleksandr Tkachenko, Ebrahim M. Songhori, Thomas Schneider, Farinaz Koushanfar. 2018

554 citations recorded by the source corpus.

We present Chameleon, a novel hybrid (mixed-protocol) framework for secure function evaluation (SFE) which enables two parties to jointly compute a function without disclosing their private inputs. Chameleon combines the best aspects of generic SFE protocols with the ones that are based upon additive secret sharing. In particular, the framework performs linear operations in the ring $\mathbbZ _2^l $ using additively secret shared values and nonlinear operations using Yao's Garbled Circuits or the Goldreich-Micali-Wigderson protocol. Chameleon departs from the common assumption of additive or linear secret sharing models where three or more parties need to communicate in the online phase: the framework allows two parties with private inputs to communicate in the online phase under the assumption of a third node generating correlated randomness in an offline phase. Almost all of the heavy cryptographic operations are precomputed in an offline phase which substantially reduces the communication overhead. Chameleon is both scalable and significantly more efficient than the ABY framework (NDSS'15) it is based on. Our framework supports signed fixed-point numbers. In particular, Chameleon's vector dot product of signed fixed-point numbers improves the efficiency of mining and classification of encrypted data for algorithms based upon heavy matrix multiplications. Our evaluation of Chameleon on a 5 layer convolutional deep neural network shows 133x and 4.2x faster executions than Microsoft CryptoNets (ICML'16) and MiniONN (CCS'17), respectively.

Subfields: Artificial Intelligence, Computational Theory and Mathematics, 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 Computational Theory and Mathematics

In Privacy & Security

Citations in the corpus

Privacy in Deep Learning: A Survey (2020, methodology): “ARDEN [106] # # # MNIST, CIFAR-10, SVHN Image Classification w/ DNN Cloak [107] # # CIFAR-100, CelebA, UTKFace Image Classification w/ DNN Cryptonets [108] # # # MNIST Image Classification w/ DNN Private Classification [109] # # # MNIST Image Classification w/ DNN TAPAS [110] # # # MNIST, Faces, Can”
Secure Evaluation of Quantized Neural Networks (2019, methodology): “Chameleon [55], which is an extension of the ABY framework by Demmler et al.”
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): “The resulting protocols can enable multiple parties to collaboratively compute an agreed-upon function without leaking input information from any party except for what can be inferred from the output [e.g., 23, 43, 95].”
Gazelle: A Low Latency Framework for Secure Neural Network Inference (2018, methodology): “However even for these the network our efﬁcient FC and Conv implementation allows us roughly 30 × and 17 × lower runtime when compared with [29] and [26] respectively.”
Privacy-preserving Machine Learning as a Service (2018, neutral): “For a batch input with size 8192, the amount of transferred data and the running time in our solution remains the same, however in [39], they increase drastically and make the solution inefficient for large input size.”
Falcon: Honest-Majority Maliciously Secure Framework for Private Deep Learning (2020, methodology): “Our evaluation set-up uses similar as compared to prior work [4, 5, 8, 11, 12].”
New Primitives for Actively-Secure MPC over Rings with Applications to Private Machine Learning (2019, methodology): “For our applications to decision tree and SVM evaluation, the most relevant are the works by de Cock et al. [28], Demmler et al. [26] and Makri et al. [29].”
Cheetah: Optimizing and Accelerating Homomorphic Encryption for Private Inference (2020, methodology): “Techniques can be categorized into two groups: HE only [11], [29], [32], [55], or multiparty computation (MPC)- based [33], [37], [50], [51].”
ABY2.0: Improved Mixed-Protocol Secure Two-Party Computation (2020, methodology): “Mixed 2PC protocols that combine GC-based and SS-based approaches benefit from their respective advantages and were used in many privacypreserving applications such as face recognition [49], fingerprint recognition [24], biometric matching [39], and machine learning [57, 73, 75, 91].”
Low Latency Privacy Preserving Inference (2018, methodology): “For example, the Chameleon system (Riazi et al., 2018) uses MPC to demonstrate private predictions on MNIST and Juvekar et al.”
Trident: Efficient 4PC Framework for Privacy Preserving Machine Learning (2019, methodology): “Similar setup for the three-party case can be found in [4], [5], [11], [12], [25].”
A Survey on Federated Learning for Resource-Constrained IoT Devices (2021, unknown): “The protocol outcome enables multiple data owners to collaboratively agree to generate a function without leaking any information [130]–[132].”
Secure Outsourced Matrix Computation and Application to Neural Networks (2018, unknown): “[42] designed Chameleon, which relies on a trusted third-party.”
TAPAS: Tricks to Accelerate (encrypted) Prediction As a Service (2018, methodology): “It is important to distinguish between approaches purely based on homomorphic encryption (described above), and those involving MultiParty Computation (MPC) techniques, such as (Mohassel & Zhang, 2017; Liu et al., 2017; Rouhani et al., 2017; Riazi et al., 2017; Chase et al.; Juvekar et al., 2018).”
Security and Privacy Issues in Deep Learning (2018, background): “] Orlandi et al. [100] Shokri and Shmatikov [29] Liu et al. [101] Liu et al. [102] Mohassel and Zhang [103] Rouhani et al. [104] Juvekar et al. [105] Chase et al. [106] Acar et al. [107] Riazi et al. [108] Aono et al. [30] III. SECUREAI Deep learning is applied to various ﬁelds ranging from autonomo”