Scalable zkSNARKs for Matrix Computations: A Generic Framework for Verifiable Deep Learning
Published in International Conference on the Theory and Application of Cryptology and Information Security (ASIACRYPT 2025), 2025
Recommended citation: Mingshu Cong, Sherman S. M. Chow, Siu-Ming Yiu, and Tsz Hon Yuen. “Scalable zkSNARKs for Matrix Computations: A Generic Framework for Verifiable Deep Learning.” In Advances in Cryptology – ASIACRYPT 2025, Lecture Notes in Computer Science, vol. 16249, pp. 363–395, 2026. https://doi.org/10.1007/978-981-95-5116-3_12
This paper presents a generic zkSNARK framework for verifiable deep learning based on matrix computations. It achieves efficient proof composition, logarithmic proof size and verification time, and architecture privacy for heterogeneous neural-network models.