From LUT to Layer: Neural Network Inference with CKKS Functional Bootstrapping

• Matheus de O. Saldanha UNICAMP https://orcid.org/0009-0000-0616-5680
• Guilherme A. Horn UNICAMP https://orcid.org/0009-0009-8296-5933
• Mauricio L. Ferreira UNICAMP https://orcid.org/0009-0002-3550-9002
• Andreis G. M. Purim UNICAMP https://orcid.org/0009-0001-5908-3213
• Alex L. D. Cassinelli UNICAMP https://orcid.org/0009-0004-3492-1796
• Hilder V. L. Pereira UNICAMP https://orcid.org/0000-0003-1303-3760

DOI: https://doi.org/10.5753/sbseg.2026.27731

Resumo

Neural networks implemented with homomorphic cryptography usually rely on costly polynomial approximations for non-linear activation functions. This work presents, to the best of our knowledge, the first implementation of neural-network inference using the recently proposed CKKS functional bootstrapping framework for arbitrary Look-Up Tables (LUTs). We implement fully connected neural networks for MNIST classification using LUT-based functional bootstrapping of sign, Heaviside, and ReLU activations, are compare against the traditional Chebyshev polynomial approximation. Across all evaluated configurations, the functional pipeline reduces inference latency by approximately 2 to 3.5 times while preserving accuracy more effectively for discontinuous activations at larger hidden widths, at the cost of an average memory overhead of approximately 1.6 times.

Referências

Al Badawi, A., Bates, J., Bergamaschi, F., Cousins, D. B., Erabelli, S., Genise, N., Halevi, S., Hunt, H., Kim, A., Lee, Y., Liu, Z., Micciancio, D., Quah, I., Polyakov, Y., R. V., S., Rohloff, K., Saylor, J., Suponitsky, D., Triplett, M., Vaikuntanathan, V., and Zucca, V. (2022). OpenFHE: Open-source fully homomorphic encryption library. Cryptology ePrint Archive. Report 2022/915, [link].

Alexandru, A., Kim, A., and Polyakov, Y. (2025). General functional bootstrapping using ckks. In Advances in Cryptology – CRYPTO 2025, volume 16002 of Lecture Notes in Computer Science, pages 304–337. Springer.

Bourse, F., Minelli, M., Minihold, M., and Paillier, P. (2018). Fast homomorphic evaluation of deep discretized neural networks. In Advances in Cryptology – CRYPTO 2018, volume 10993 of Lecture Notes in Computer Science, pages 483–512. Springer.

Cheon, J. H., Kim, A., Kim, M., and Song, Y. (2017). Homomorphic encryption for arithmetic of approximate numbers. In Advances in Cryptology – ASIACRYPT 2017, volume 10624 of Lecture Notes in Computer Science, pages 409–437. Springer.

Chillotti, I., Gama, N., Georgieva, M., and Izabachène, M. (2020). Tfhe: Fast fully homomorphic encryption over the torus: I. chillotti et al. Journal of Cryptology, 33(1):34–91.

Fan, J. and Vercauteren, F. (2012). Somewhat practical fully homomorphic encryption. Cryptology ePrint Archive, Report 2012/144.

Gentry, C. (2009). A Fully Homomorphic Encryption Scheme. PhD thesis, Stanford University.

Gilad-Bachrach, R., Dowlin, N., Laine, K., Lauter, K., Naehrig, M., and Wernsing, J. (2016). Cryptonets: Applying neural networks to encrypted data with high throughput and accuracy. In Balcan, M. F. and Weinberger, K. Q., editors, Proceedings of the 33rd International Conference on Machine Learning, volume 48 of Proceedings of Machine Learning Research, pages 201–210, New York, NY, USA. PMLR.

Halevi, S. and Shoup, V. (2014). Algorithms in HElib. In Advances in Cryptology – CRYPTO 2014, pages 554–571. Springer.

Hesamifard, E., Takabi, H., and Ghasemi, M. (2016). Cryptodl: Towards deep learning over encrypted data. In Proceedings of the Annual Computer Security Applications Conference (ACSAC), Los Angeles, CA, USA.

Liu, B., Ding, M., Shaham, S., Rahayu, W., Farokhi, F., and Lin, Z. (2021). When machine learning meets privacy: A survey and outlook. ACM Computing Surveys (CSUR), 54(2):1–36.

Pulido-Gaytan, L. B., Tchernykh, A., Cortés-Mendoza, J. M., Babenko, M., and Radchenko, G. (2020). A survey on privacy-preserving machine learning with fully homomorphic encryption. In Latin American High Performance Computing Conference, pages 115–129. Springer.

Yang, Z., Niu, C., Wei, B., Huang, Z., Hong, C., and Wei, T. (2025). Rboot: Accelerating homomorphic neural network inference by fusing relu within bootstrapping. Cryptology ePrint Archive, Report 2025/1534.