Eurocrypt 2024 witnessed significant strides in cryptographic research, with Keita Xagawa of the CRC-TII Post-quantum team presenting groundbreaking work on memory-tight reductions for post-quantum signature schemes. By exploring the multi-challenge security of signature schemes derived from lossy identification within the quantum random oracle model, Xagawa's research provides crucial insights into fortifying security against quantum adversaries. Additionally, the paper sheds light on signature schemes from preimage-sampleable functions, showcasing their potential for heightened security through memory-tight reductions.
In a parallel domain, Elena Kirshanova and Andre Esser from the CRC-TII’s Cryptanalysis team alongside Léo Ducas and Simona Etinski, investigated the adaptation of sieving techniques from lattices to codes, as proposed by Guo, Johansson, and Nguyen in Eprint'23. Their research offers an asymptotic analysis and suggests a systematic use of Locality Sensitive Hashing and Filtering (LSH/F) to enhance the efficiency of sieving. Their work establishes a baseline decoding complexity for sieving with conventional worst parameters, surpassing previous algorithms and offering insights into practical applications of the Random-Spherical-Code-Product (RPC) and an alternative specific to the Hamming Sphere.
Meanwhile,addressing the security challenges surrounding Deep Neural Networks (DNNs), Isaac A. Canales-Martínez, Jorge Chavez-Saab, Anna Hambitzer, Francisco Rodríguez-Henríquez, Nitin Satpute and Adi Shamir, delve into efficient parameter extraction techniques. Focusing on ReLU-based DNNs, the authors present novel approaches to reduce the time required for parameter extraction, marking a substantial improvement over previous methods. Their work demonstrates remarkable efficiency, enabling the extraction of millions of parameters from real-world image-processing networks in just 30 minutes on a standard computer, thus enhancing security for a broad range of networks.
Check out the full articles below:
Signatures with Memory-Tight Security in the Quantum Random Oracle Model by Keita Xagawa // CRC-TII, Post-quantum team
https://eprint.iacr.org/2023/1734
Asymptotics and Improvements of Sieving for Codes by Léo Ducas, Simona Etinski, Andre Esser, Elena Kirshanova // Andre and Elena work for CRC-TII, Cryptanalysis team
https://eprint.iacr.org/2023/1577
Polynomial Time Cryptanalytic Extraction of Neural Network Models by Isaac A. Canales-Martínez, Jorge Chavez-Saab, Anna Hambitzer, Francisco Rodríguez-Henríquez, Nitin Satpute of TII and Adi Shamir of the Weizmann Institute of Science
https://eprint.iacr.org/2023/1526