Algorithms in HElib

By Shai Halevi, Victor Shoup. Annual International Cryptology Conference, 2014

734 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computational Theory and Mathematics

View Paper: https://www.semanticscholar.org/paper/b7eefb9f25518c9193f40617f4a7d8cbbe36d792

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

1. Kim Laine
2. Hao Chen
3. Miran Kim
4. Wei Dai
5. Shai Halevi
6. Yongsoo Song
7. Michael Naehrig
8. Chris Peikert

In Computational Theory and Mathematics

1. Kim Laine
2. Hao Chen
3. Shai Halevi
4. Miran Kim
5. Kristin E. Lauter
6. Wei Dai
7. Michael Naehrig
8. Chris Peikert

Citations in the corpus

• Gazelle: A Low Latency Framework for Secure Neural Network Inference (2018, methodology): “The diagonal method as described in the work of Halevi and Shoup [20] (and implemented in [19]) provides another potential solution to the problem of a large number of output ciphertexts.”
• Accelerating Homomorphic Evaluation on Reconfigurable Hardware (2015, methodology): “A software library that implements the Brakerski-Gentry-Vaikuntanathan (BGV) [7,8] scheme is described in [27].”
• Bootstrapping for Approximate Homomorphic Encryption (2018, methodology): “As discussed in [32], it can be efficiently done by handling the matrix in a diagonal order and making use of SIMD computation.”
• Improving TFHE: faster packed homomorphic operations and efficient circuit bootstrapping (2017, methodology): “Some public implementations are available, namely Helib [22], FV-NFlib [24] and SEAL [29], based on BGV [5, 18], and FHEW [17] and TFHE [14], based on GSW [21, 17, 12].”
• FHEW: Bootstrapping Homomorphic Encryption in Less Than a Second (2015, methodology): “In fact, using complex ciphertexts packing and homomorphic SIMD techniques, [23] achieves an amortized cost (per homomorphic bit operation) which we estimate to be in the same order of magnitude as our solution.”
• CryptoDL: Deep Neural Networks over Encrypted Data (2017, methodology): “We used HELib [14] for implementation and all computations were run on a computer with 16GB RAM, Intel Xeon E5-2640, 2.”
• Secure Outsourced Matrix Computation and Application to Neural Networks (2018, positive): “Halevi and Shoup [25] introduced a matrix encoding method based on its diagonal decomposition, putting the matrix in diagonal order and mapping each of them to a single ciphertext.”
• Privacy-Preserving Classification on Deep Neural Network (2017, methodology): “4In addition, HElib [26], a software library that implements this homomorphic encryption scheme, is available on GitHub [25].”
• Simple Encrypted Arithmetic Library - SEAL v2.1 (2016, background): “One reason for this is that the documentation available for HElib [24, 25, 26] does not in general make clear how to select optimal parameters for performance, and in [26, Appendix A] it is noted ‘[t]he BGV implementation in HElib relies on a myriad of parameters .”
• Efficient Multi-Key Homomorphic Encryption with Packed Ciphertexts with Application to Oblivious Neural Network Inference (2019, methodology): “We start with a useful aggregation operation across plaintext slots from the literature [31, 38].”
• F1: A Fast and Programmable Accelerator for Fully Homomorphic Encryption (2021, methodology): “Despite these limitations, prior work has devised reasonably efficient implementations of key algorithms, including linear algebra [38], neural network inference [15, 36], logistic regression [39], and genome processing [11].”
• Cheetah: Lean and Fast Secure Two-Party Deep Neural Network Inference (2022, methodology): “FC [27] O(noni) O(n̄(n∗+ log2( no n∗ ))) Cheetah O(noni) 0”
• Simple Encrypted Arithmetic Library-SEAL (2017, background): “One reason for this is that the documentation available for HElib [24, 25, 26] does not in general make clear how to select optimal parameters for performance, and in [26, Appendix A] it is noted ‘[t]he BGV implementation in HElib relies on a myriad of parameters ... it takes some experimentation to”
• A Fast Single Server Private Information Retrieval Protocol with Low Communication Cost (2014, methodology): “The implementation is based on HElib [36], an open source implementation of the BGV FHE scheme.”
• Preserving Multi-party Machine Learning with Homomorphic Encryption (2016, methodology): “For generating schemes in HELib, we set some values for input parameters, k = 80, s = 1, w = 64 (for details about these parameters see (7)).”
• Polynomial multipliers for fully homomorphic encryption on FPGA (2015, methodology): “In Section V, we compare and discuss the proposed solutions and we compare to software runtimes.”