ENSEI: Efficient Secure Inference via Frequency-Domain Homomorphic Convolution for Privacy-Preserving Visual Recognition.
Song BianTianchen WangMasayuki HiromotoYiyu ShiTakashi SatoPublished in: CVPR (2020)
Keyphrases
- privacy preserving
- frequency domain
- visual recognition
- vertically partitioned data
- scalar product
- fourier transform
- horizontally partitioned data
- secure multiparty computation
- multi party
- privacy preserving data mining
- fast fourier transform
- spatial domain
- sensitive data
- discrete fourier transform
- fourier domain
- privacy preservation
- image classification
- feature extraction
- homomorphic encryption
- user privacy
- sensitive information
- privacy protection
- privacy concerns
- denoising
- horizontally partitioned
- privacy guarantees
- object recognition
- private information
- data privacy
- subband
- privacy sensitive
- image processing
- differential privacy
- privacy preserving association rule mining
- image quality
- machine learning
- semi honest