{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,11,19]],"date-time":"2025-11-19T15:49:15Z","timestamp":1763567355774,"version":"3.45.0"},"reference-count":43,"publisher":"Springer Science and Business Media LLC","issue":"9","license":[{"start":{"date-parts":[[2025,10,16]],"date-time":"2025-10-16T00:00:00Z","timestamp":1760572800000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2025,10,16]],"date-time":"2025-10-16T00:00:00Z","timestamp":1760572800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["J. King Saud Univ. Comput. Inf. Sci."],"published-print":{"date-parts":[[2025,11]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>While fully homomorphic encryption (FHE) provides strong security guarantees for privacy-preserving machine learning, its practical adoption is hindered by substantial computational overhead where bootstrapping operations dominate 84% of inference latency. Existing single-image approaches further suffer from significant ciphertext slot underutilization in convolutional layers. To address these limitations, we propose a novel multi-image homomorphic convolution framework featuring three key innovations: (1) simultaneous multi-image packing enabling batched convolution at single-image computational cost, (2) complex number encoding utilizing both real and imaginary components to double slot utilization, and (3) sparsity-aware ciphertext merging that dynamically consolidates underutilized slots. Our comprehensive evaluation on CIFAR-10, CIFAR-100 and ImageNet using ResNet architectures demonstrates that processing 16 images concurrently achieves 74% latency reduction compared to single-image baselines while maintaining a 30% speed advantage over multi-image variants of prior work with negligible accuracy degradation (&lt;\u20090.7% drop). These results establish a new state-of-the-art for efficient and scalable privacy-preserving inference.<\/jats:p>","DOI":"10.1007\/s44443-025-00290-1","type":"journal-article","created":{"date-parts":[[2025,10,16]],"date-time":"2025-10-16T14:17:37Z","timestamp":1760624257000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Optimization of CNN inference for multi-image with fully homomorphic encryption"],"prefix":"10.1007","volume":"37","author":[{"ORCID":"https:\/\/orcid.org\/0009-0007-1202-0198","authenticated-orcid":false,"given":"Chao","family":"Xu","sequence":"first","affiliation":[]},{"given":"Yun","family":"Wang","sequence":"additional","affiliation":[]}],"member":"297","published-online":{"date-parts":[[2025,10,16]]},"reference":[{"key":"290_CR1","doi-asserted-by":"crossref","unstructured":"Albrecht M, Chase M, Chen H, Ding J, Goldwasser S, Gorbunov S, ... Lauter K (2022) Homomorphic encryption standard. In: Protecting privacy through homomorphic encryption, Springer, pp 31\u201362. https:\/\/eprint.iacr.org\/2019\/939.pdf. Accessed May 2025","DOI":"10.1007\/978-3-030-77287-1_2"},{"key":"290_CR2","doi-asserted-by":"publisher","unstructured":"Bertolace A, Gatsis K, Margellos K (2023) Homomorphically encrypted gradient descent algorithms for quadratic programming. Paper presented at the 2023 62nd IEEE Conference on Decision and Control (CDC). https:\/\/doi.org\/10.1109\/CDC49753.2023.10383503","DOI":"10.1109\/CDC49753.2023.10383503"},{"key":"290_CR3","doi-asserted-by":"publisher","unstructured":"Boemer F, Costache A, Cammarota R, Wierzynski C (2019) nGraph-HE2: A high-throughput framework for neural network inference on encrypted data. Paper presented at the Proceedings of the 7th ACM workshop on encrypted computing & applied homomorphic cryptography. https:\/\/doi.org\/10.1145\/3338469.3358944","DOI":"10.1145\/3338469.3358944"},{"key":"290_CR4","doi-asserted-by":"publisher","unstructured":"Boemer F, Lao Y, Cammarota R, Wierzynski C (2019) nGraph-HE: a graph compiler for deep learning on homomorphically encrypted data. Paper presented at the Proceedings of the 16th ACM international conference on computing frontiers. https:\/\/doi.org\/10.1145\/3310273.3323047","DOI":"10.1145\/3310273.3323047"},{"key":"290_CR5","doi-asserted-by":"publisher","unstructured":"Bossuat J-P, Mouchet C, Troncoso-Pastoriza J, Hubaux J-P (2021) Efficient bootstrapping for approximate homomorphic encryption with non-sparse keys. Paper presented at the Annual International Conference on the Theory and Applications of Cryptographic Techniques. https:\/\/doi.org\/10.1007\/978-3-030-77870-5_21","DOI":"10.1007\/978-3-030-77870-5_21"},{"key":"290_CR6","doi-asserted-by":"publisher","unstructured":"Brakerski Z, Vaikuntanathan VJSJOC (2014) Efficient fully homomorphic encryption from (standard) LWE. 43(2):831\u2013871. https:\/\/doi.org\/10.1137\/120868669","DOI":"10.1137\/120868669"},{"key":"290_CR7","doi-asserted-by":"publisher","unstructured":"Brutzkus A, Gilad-Bachrach R, Elisha O (2019) Low latency privacy preserving inference. Paper presented at the International Conference on Machine Learning. https:\/\/doi.org\/10.48550\/arXiv.1812.10659","DOI":"10.48550\/arXiv.1812.10659"},{"key":"290_CR8","doi-asserted-by":"publisher","unstructured":"Chen H, Chillotti I, Song Y (2019) Improved bootstrapping for approximate homomorphic encryption. Paper presented at the Annual International Conference on the Theory and Applications of Cryptographic Techniques. https:\/\/doi.org\/10.1007\/978-3-030-17656-3_2","DOI":"10.1007\/978-3-030-17656-3_2"},{"key":"290_CR9","doi-asserted-by":"publisher","first-page":"89497","DOI":"10.1109\/ACCESS.2019.2925425","volume":"7","author":"JH Cheon","year":"2019","unstructured":"Cheon JH, Hhan M, Hong S, Son YJIA (2019) A hybrid of dual and meet-in-the-middle attack on sparse and ternary secret LWE. IEEE Access 7:89497\u201389506. https:\/\/doi.org\/10.1109\/ACCESS.2019.2925425","journal-title":"IEEE Access"},{"key":"290_CR10","doi-asserted-by":"publisher","unstructured":"Cheon JH, Kim A, Kim M, Song Y (2017) Homomorphic encryption for arithmetic of approximate numbers. Paper presented at the International conference on the theory and application of cryptology and information security. https:\/\/doi.org\/10.1007\/978-3-319-70694-8_15","DOI":"10.1007\/978-3-319-70694-8_15"},{"key":"290_CR11","doi-asserted-by":"publisher","unstructured":"Cheon JH, Han K, Kim A, Kim M, Song Y (2018a) Bootstrapping for approximate homomorphic encryption. Paper presented at the Annual International Conference on the Theory and Applications of Cryptographic Techniques. https:\/\/doi.org\/10.1007\/978-3-319-78381-9_14","DOI":"10.1007\/978-3-319-78381-9_14"},{"key":"290_CR12","doi-asserted-by":"publisher","unstructured":"Cheon JH, Han K, Kim A, Kim M, Song Y (2018b) A full RNS variant of approximate homomorphic encryption. Paper presented at the International Conference on Selected Areas in Cryptography. https:\/\/doi.org\/10.1007\/978-3-030-10970-7_16","DOI":"10.1007\/978-3-030-10970-7_16"},{"key":"290_CR13","doi-asserted-by":"publisher","unstructured":"Cheon JH, Kim D, Kim D (2020) Efficient homomorphic comparison methods with optimal complexity. Paper presented at the International Conference on the Theory and Application of Cryptology and Information Security. https:\/\/doi.org\/10.1007\/978-3-030-64834-3_8","DOI":"10.1007\/978-3-030-64834-3_8"},{"key":"290_CR14","doi-asserted-by":"publisher","unstructured":"Chillotti I, Joye M, Paillier P (2021) Programmable bootstrapping enables efficient homomorphic inference of deep neural networks. Paper presented at the international symposium on cyber security cryptography and machine learning. https:\/\/doi.org\/10.1007\/978-3-030-78086-9_1","DOI":"10.1007\/978-3-030-78086-9_1"},{"key":"290_CR15","doi-asserted-by":"publisher","unstructured":"Chou E, Beal J, Levy D, Yeung S, Haque A, Fei-Fei LJAPA (2018) Faster cryptonets: leveraging sparsity for real-world encrypted inference. https:\/\/doi.org\/10.48550\/arXiv.1811.09953","DOI":"10.48550\/arXiv.1811.09953"},{"key":"290_CR16","doi-asserted-by":"publisher","unstructured":"Dathathri R, Saarikivi O, Chen H, Laine K, Lauter K, Maleki S, . . . Mytkowicz T (2019) CHET: an optimizing compiler for fully-homomorphic neural-network inferencing. Paper presented at the Proceedings of the 40th ACM SIGPLAN conference on programming language design and implementation. https:\/\/doi.org\/10.1145\/3314221.3314628","DOI":"10.1145\/3314221.3314628"},{"key":"290_CR17","doi-asserted-by":"publisher","unstructured":"Dathathri R, Kostova B, Saarikivi O, Dai W, Laine K, Musuvathi M (2020) EVA: an encrypted vector arithmetic language and compiler for efficient homomorphic computation. Paper presented at the Proceedings of the 41st ACM SIGPLAN conference on programming language design and implementation. https:\/\/doi.org\/10.1145\/3385412.3386023","DOI":"10.1145\/3385412.3386023"},{"key":"290_CR18","doi-asserted-by":"publisher","unstructured":"Fredrikson M, Jha S, Ristenpart T (2015) Model inversion attacks that exploit confidence information and basic countermeasures. Paper presented at the Proceedings of the 22nd ACM SIGSAC conference on computer and communications security. https:\/\/doi.org\/10.1145\/2810103.2813677","DOI":"10.1145\/2810103.2813677"},{"key":"290_CR19","doi-asserted-by":"publisher","DOI":"10.1109\/JIOT.2025.3540097","author":"S Gao","year":"2025","unstructured":"Gao S, Zhang Z, Iu HH-C, Ding S, Mou J, Erkan U, Toktas A, Li Qi, Wang C, Cao Y, Cao YJ (2025b) A parallel color image encryption algorithm based on a 2D logistic-rulkov neuron map. IEEE Internet Things J. https:\/\/doi.org\/10.1109\/JIOT.2025.3540097","journal-title":"IEEE Internet Things J"},{"key":"290_CR20","doi-asserted-by":"publisher","unstructured":"Gao S, Iu HH-C, Mou J, Erkan U, Liu J, Wu R, . . . Fractals (2024) Temporal action segmentation for video encryption. 183:114958. https:\/\/doi.org\/10.1016\/j.chaos.2024.114958","DOI":"10.1016\/j.chaos.2024.114958"},{"key":"290_CR21","doi-asserted-by":"publisher","unstructured":"Gao S, Iu HH-C, Erkan U, Simsek C, Toktas A, Cao Y, . . . Technology SFV (2025) A 3D memristive cubic map with dual discrete memristors: design, implementation, and application in image encryption. https:\/\/doi.org\/10.1109\/TCSVT.2025.3545868","DOI":"10.1109\/TCSVT.2025.3545868"},{"key":"290_CR22","unstructured":"Gentry C (2009) A fully homomorphic encryption scheme: Stanford university. https:\/\/evervault.com\/papers\/gentry.pdf. Accessed May 2025"},{"key":"290_CR23","doi-asserted-by":"publisher","unstructured":"Gilad-Bachrach R, Dowlin N, Laine K, Lauter K, Naehrig M, Wernsing J (2016) Cryptonets: applying neural networks to encrypted data with high throughput and accuracy. Paper presented at the International conference on machine learning. https:\/\/doi.org\/10.1109\/RWS.2014.6830099","DOI":"10.1109\/RWS.2014.6830099"},{"key":"290_CR24","doi-asserted-by":"publisher","unstructured":"Han K, Ki D (2020) Better bootstrapping for approximate homomorphic encryption. Paper presented at the cryptographers\u2019 track at the RSA conference. https:\/\/doi.org\/10.1007\/978-3-030-40186-3_16","DOI":"10.1007\/978-3-030-40186-3_16"},{"key":"290_CR25","doi-asserted-by":"publisher","unstructured":"Hesamifard E, Takabi H, Ghasemi MJAPA (2017) Cryptodl: deep neural networks over encrypted data. https:\/\/doi.org\/10.48550\/arXiv.1711.05189","DOI":"10.48550\/arXiv.1711.05189"},{"key":"290_CR26","doi-asserted-by":"publisher","unstructured":"Jung W, Kim S, Ahn JH, Cheon JH, Lee YJITOCH, Systems E (2021) Over 100x faster bootstrapping in fully homomorphic encryption through memory-centric optimization with GPUs. 114\u2013148. https:\/\/doi.org\/10.46586\/tches.v2021.i4.114-148","DOI":"10.46586\/tches.v2021.i4.114-148"},{"key":"290_CR27","unstructured":"Juvekar C, Vaikuntanathan V, Chandrakasan A (2018) {GAZELLE}: A low latency framework for secure neural network inference. In: Paper presented at the 27th USENIX security symposium (USENIX security 18). https:\/\/www.usenix.org\/conference\/usenixsecurity18\/presentation\/juvekar. Accessed May 2025"},{"key":"290_CR28","doi-asserted-by":"publisher","first-page":"2175","DOI":"10.1109\/TIFS.2023.3263631","volume":"18","author":"D Kim","year":"2023","unstructured":"Kim D, Guyot C (2023) Optimized privacy-preserving CNN inference with fully homomorphic encryption. IEEE Trans Inf Forensics Secur 18:2175\u20132187. https:\/\/doi.org\/10.1109\/TIFS.2023.3263631","journal-title":"IEEE Trans Inf Forensics Secur"},{"key":"290_CR29","unstructured":"Krizhevsky A, Hinton G (2009) Learning multiple layers of features from tiny images. http:\/\/www.cs.utoronto.ca\/~kriz\/learning-features-2009-TR.pdf. Accessed May 2025"},{"key":"290_CR30","unstructured":"LeCun, YJHYlCEM (1998) The MNIST database of handwritten digits. http:\/\/yann.lecun.com\/exdb\/mnist. Accessed May 2025"},{"key":"290_CR31","doi-asserted-by":"publisher","first-page":"26163","DOI":"10.1109\/ACCESS.2022.3155882","volume":"10","author":"E Lee","year":"2022","unstructured":"Lee E, Lee J-W, Kim Y-S, No J-S (2022a) Optimization of homomorphic comparison algorithm on RNS-CKKS scheme. J i a 10:26163\u201326176. https:\/\/doi.org\/10.1109\/ACCESS.2022.3155882","journal-title":"J i a"},{"key":"290_CR32","doi-asserted-by":"publisher","unstructured":"Lee E, Lee J-W, No J-S, Kim Y-S JITOD, Computing S (2021a) Minimax approximation of sign function by composite polynomial for homomorphic comparison. 19(6):3711\u20133727. https:\/\/doi.org\/10.1109\/TDSC.2021.3105111","DOI":"10.1109\/TDSC.2021.3105111"},{"key":"290_CR33","doi-asserted-by":"publisher","unstructured":"Lee J-W, Kang H, Lee Y, Choi W, Eom J, Deryabin M, ... Kim Y-S (2021b) Privacy-preserving machine learning with fully homomorphic encryption for deep neural network. https:\/\/doi.org\/10.1109\/ACCESS.2022.3159694","DOI":"10.1109\/ACCESS.2022.3159694"},{"key":"290_CR34","unstructured":"Lee E, Lee J-W, Lee J, Kim Y-S, Kim Y, No J-S, Choi W (2022b) Low-complexity deep convolutional neural networks on fully homomorphic encryption using multiplexed parallel convolutions. In: Paper presented at the International Conference on Machine Learning. https:\/\/eprint.iacr.org\/2021\/1688. Accessed May 2025"},{"key":"290_CR35","doi-asserted-by":"publisher","unstructured":"Lee J, Lee E, Lee J-W, Kim Y, Kim Y-S, No J-SJIA (2023) Precise approximation of convolutional neural networks for homomorphically encrypted data. 11:62062-62076. https:\/\/doi.org\/10.1007\/978-3-030-64834-3_8","DOI":"10.1007\/978-3-030-64834-3_8"},{"key":"290_CR36","doi-asserted-by":"publisher","unstructured":"Li Q, Wang X, Ma B, Wang X, Wang C, Gao S, . . . Technology SFV (2021) Concealed attack for robust watermarking based on generative model and perceptual loss. 32(8):5695\u20135706. https:\/\/doi.org\/10.1109\/TCSVT.2021.3138795","DOI":"10.1109\/TCSVT.2021.3138795"},{"key":"290_CR37","doi-asserted-by":"publisher","unstructured":"Li Q, Ma B, Wang X, Wang C, Gao SJITOC, Briefs SIE (2023) Image steganography in color conversion. 71(1):106\u2013110. https:\/\/doi.org\/10.1109\/TCSII.2023.3300330","DOI":"10.1109\/TCSII.2023.3300330"},{"key":"290_CR38","doi-asserted-by":"publisher","unstructured":"Lin Y, Xie Z, Chen T, Cheng X, Wen HJESWA (2024) Image privacy protection scheme based on high-quality reconstruction DCT compression and nonlinear dynamics. 257:124891. https:\/\/doi.org\/10.1016\/j.eswa.2024.124891","DOI":"10.1016\/j.eswa.2024.124891"},{"key":"290_CR39","doi-asserted-by":"publisher","unstructured":"Lou Q, Jiang L (2021) Hemet: A homomorphic-encryption-friendly privacy-preserving mobile neural network architecture. Paper presented at the International conference on machine learning. https:\/\/doi.org\/10.48550\/arXiv.2106.00038","DOI":"10.48550\/arXiv.2106.00038"},{"issue":"3","key":"290_CR40","doi-asserted-by":"publisher","first-page":"211","DOI":"10.1007\/s11263-015-0816-y","volume":"115","author":"O Russakovsky","year":"2015","unstructured":"Russakovsky O, Deng J, Su H, Krause J, Satheesh S, Ma S, Huang Z, Karpathy A, Khosla A, Bernstein M, Berg AC, Fei-Fei Li, Bernstein MJI (2015) Imagenet large scale visual recognition challenge. Int J Comput vis 115(3):211\u2013252. https:\/\/doi.org\/10.1007\/s11263-015-0816-y","journal-title":"Int J Comput vis"},{"issue":"10","key":"290_CR41","doi-asserted-by":"publisher","DOI":"10.3390\/app13106117","volume":"13","author":"C Song","year":"2023","unstructured":"Song C, Huang RJAS (2023) Secure convolution neural network inference based on homomorphic encryption. Appl Sci 13(10):6117. https:\/\/doi.org\/10.3390\/app13106117","journal-title":"Appl Sci"},{"key":"290_CR42","doi-asserted-by":"publisher","unstructured":"Wang H, Cheng X, Wu H, Luo X, Ma B, Zong H, . . . Representation I (2025) A GAN-based anti-forensics method by modifying the quantization table in JPEG header file. 104462. https:\/\/doi.org\/10.1016\/j.jvcir.2025.104462","DOI":"10.1016\/j.jvcir.2025.104462"},{"key":"290_CR43","doi-asserted-by":"publisher","unstructured":"Wang R, Zeng Q, Yang Z, Zhang PJTJOS (2025) Cloud-based secure human action recognition with fully homomorphic encryption. 81(1):12. https:\/\/doi.org\/10.1007\/s11227-024-06512-z","DOI":"10.1007\/s11227-024-06512-z"}],"container-title":["Journal of King Saud University Computer and Information Sciences"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s44443-025-00290-1.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/article\/10.1007\/s44443-025-00290-1\/fulltext.html","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s44443-025-00290-1.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,11,19]],"date-time":"2025-11-19T15:45:25Z","timestamp":1763567125000},"score":1,"resource":{"primary":{"URL":"https:\/\/link.springer.com\/10.1007\/s44443-025-00290-1"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,10,16]]},"references-count":43,"journal-issue":{"issue":"9","published-print":{"date-parts":[[2025,11]]}},"alternative-id":["290"],"URL":"https:\/\/doi.org\/10.1007\/s44443-025-00290-1","relation":{},"ISSN":["1319-1578","2213-1248"],"issn-type":[{"type":"print","value":"1319-1578"},{"type":"electronic","value":"2213-1248"}],"subject":[],"published":{"date-parts":[[2025,10,16]]},"assertion":[{"value":"6 May 2025","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"15 September 2025","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"16 October 2025","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"All authors declare that they have no conflicts of interest. No financial or non-financial benefits have been received from any party related to the subject of this paper.","order":2,"name":"Ethics","group":{"name":"EthicsHeading","label":"Conflict of interest"}}],"article-number":"260"}}