2018/06/07 by Behrooz Razeghi, Slava Voloshynovskiy, Razeghi, Behrooz +5
Computer Science · Engineering · #Cryptography and Security (cs.CR) #Data Structures and Algorithms (cs.DS) #Databases (cs.DB) #Distributed #FOS: Computer and information sciences #Information Theory (cs.IT) #Internet Traffic Analysis and Secure E-voting #Parallel #Sparse and Compressive Sensing Techniques #Wireless Signal Modulation Classification #and Cluster Computing (cs.DC)
paper · pdf · doi:10.48550/arxiv.1806.08658
openalex publication_date 2018/06/07 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
We propose a new computationally efficient privacy-preserving identification\nframework based on layered sparse coding. The key idea of the proposed\nframework is a sparsifying transform learning with ambiguization, which\nconsists of a trained linear map, a component-wise nonlinearity and a privacy\namplification. We introduce a practical identification framework, which\nconsists of two phases: public and private identification. The public untrusted\nserver provides the fast search service based on the sparse privacy protected\ncodebook stored at its side. The private trusted server or the local client\napplication performs the refined accurate similarity search using the results\nof the public search and the layered sparse codebooks stored at its side. The\nprivate search is performed in the decoded domain and also the accuracy of\nprivate search is chosen based on the authorization level of the client. The\nefficiency of the proposed method is in computational complexity of encoding,\ndecoding, "encryption" (ambiguization) and "decryption" (purification) as well\nas storage complexity of the codebooks.\n