Paper 2018/1245

Multi-dimensional Packing for HEAAN for Approximate Matrix Arithmetics

Jung Hee Cheon, Andrey Kim, and Donggeon Yhee


HEAAN is a homomorphic encryption (HE) scheme for approximate arithmetics. Its vector packing technique proved its potential in cryptographic applications requiring approximate computations, including data analysis and machine learning. In this paper, we propose MHEAAN - a generalization of HEAAN to the case of a tensor structure of plaintext slots. Our design takes advantage of the HEAAN scheme, that the precision losses during the evaluation are limited by the depth of the circuit, and it exceeds no more than one bit compared to unencrypted approximate arithmetics, such as floating point operations. Due to the multi-dimensional structure of plaintext slots along with rotations in various dimensions, MHEAAN is a more natural choice for applications involving matrices and tensors. We provide a concrete two-dimensional construction and show the efficiency of our scheme on several matrix operations, such as matrix multiplication, matrix transposition, and inverse. As an application, we implement the non-interactive Deep Neural Network (DNN) classification algorithm on encrypted data and encrypted model. Due to our efficient bootstrapping, the implementation can be easily extended to DNN structure with an arbitrary number of hidden layers

Available format(s)
Publication info
Preprint. MINOR revision.
Homomorphic encryptionapproximate arithmeticsmatrix operationsbootstrappinglinear transformationencrypted deep neural network.
Contact author(s)
kimandrik @ snu ac kr
2019-01-03: received
Short URL
Creative Commons Attribution


      author = {Jung Hee Cheon and Andrey Kim and Donggeon Yhee},
      title = {Multi-dimensional Packing for HEAAN for Approximate Matrix Arithmetics},
      howpublished = {Cryptology ePrint Archive, Paper 2018/1245},
      year = {2018},
      note = {\url{}},
      url = {}
Note: In order to protect the privacy of readers, does not use cookies or embedded third party content.