ILIA ILIASHENKO (Илья Ильяшенко)
My research interests lie at the intersection of cryptography and computational number theory, focusing on secure computation methods. More precisely, I study homomorphic encryption schemes based on Euclidean lattices, secure multi-party computation methods, interactive proofs and their applications to data privacy and integrity.
 |
|
Selected projects
World ID iris matching
co-designer and implementer, Inversed Tech, 2024–2026.
The three-party SMPC protocol behind World ID biometric authentication, written in Rust. Irises are matched by fractional Hamming distance under ABY3, so that they stay protected both at rest and in use.
CipherCore
contributor, Ciphermode Labs, 2021–2024.
An open-source Rust implementation of the semi-honest SMPC protocol ABY3, together with SMPC solutions built on top of it for machine-learning inference, private data collaboration and private database retrieval.
FINAL
co-author, COSIC, KU Leuven, 2021–2022.
FINAL is a fully homomorphic encryption scheme that combines the LWE and NTRU problems to bootstrap faster and with smaller keys than TFHE. The linked repository is our C++ implementation.
Microsoft APSI
protocol co-designer, COSIC, KU Leuven and Microsoft Research, 2020–2021.
APSI is Microsoft's open-source library for asymmetric private set intersection. It is based on the labeled PSI protocol that we designed to cut both the computation and the communication of unbalanced PSI from homomorphic encryption.
Microsoft SEAL
research intern, Microsoft Research, 2018–2019.
SEAL is a C++ library realizing several homomorphic encryption schemes; it powers the Password Monitor in the Microsoft Edge browser. My contribution to SEAL is the implementation of the CKKS scheme for approximate arithmetic.
Horizon 2020 HEAT
contributor, COSIC, KU Leuven, 2015–2018.
HEAT (Homomorphic Encryption Applications and Technology) was an EU project on making homomorphic encryption practical. I worked on the smart grid use case, where a licensed third party forecasts the energy consumption of a set of households without ever seeing the smart meter readings. The resulting protocol evaluates a group method of data handling network under the BFV scheme to compute a full encrypted forecast.
Publications
On Polynomial Functions Modulo pe
and Faster Bootstrapping for Homomorphic Encryption
with Robin Geelen, Jiayi Kang and Frederik Vercauteren,
Proceedings of EUROCRYPT 2023.
video
FINAL: Faster FHE instantiated with NTRU and LWE
with Charlotte Bonte, Jeongeun Park, Hilder V. L. Pereira and Nigel P. Smart,
Proceedings of ASIACRYPT 2022.
video
Homomorphically counting elements with the same property
with Malika Izabachène, Axel Mertens and Hilder V. L. Pereira,
PoPETS 2022.
video
When HEAAN Meets FV: a New Somewhat Homomorphic Encryption
with Reduced Memory Overhead
with Hao Chen and Kim Laine,
Proceedings of IMA CC 2021.
slides
Integer Functions Suitable for Homomorphic Encryption over Finite Fields
with Christophe Nègre and Vincent Zucca,
Proceedings of WAHC 2021.
slides
Labeled PSI from Homomorphic Encryption
with Reduced Computation and Communication
with Kelong Cong, Radames Cruz Moreno, Mariana Botelho da Gama,
Wei Dai, Kim Laine and Michael Rosenberg,
Proceedings of ACM CCS 2021.
slides
Faster homomorphic comparison operations for BGV and BFV
with Vincent Zucca,
PoPETS 2021.
slides video
Homomorphic string search with constant multiplicative depth
with Charlotte Bonte,
Proceedings of ACM CCSW 2020.
slides video
Efficiently Processing Complex-valued Data in Homomorphic Encryption
with Carl Bootland, Wouter Castryck and Frederik Vercauteren,
Journal of Mathematical Cryptology 14(1, Special Issue MATHCRYPT 2018): pp. 55-65, 2020.
slides video
Homomorphic SIM2D Operations: Single Instruction Much More Data
with Wouter Castryck and Frederik Vercauteren,
Proceedings of EUROCRYPT 2018.
slides
Faster Homomorphic Function Evaluation Using Non-integral Base Encoding
with Charlotte Bonte, Carl Bootland, Joppe W. Bos, Wouter Castryck and Frederik Vercauteren,
Proceedings of CHES 2017.
slides video
Privacy-friendly Forecasting for the Smart Grid Using Homomorphic Encryption and the Group Method of Data Handling
with Joppe W. Bos, Wouter Castryck and Frederik Vercauteren,
Proceedings of AFRICACRYPT 2017.
slides
On Error Distributions in Ring-based LWE
with Wouter Castryck and Frederik Vercauteren,
Proceedings of ANTS XII,
LMS Journal of Computation and Mathematics 19, A, pp. 130-145, 2016.
slides
Provably Weak Instances of Ring-LWE Revisited
with Wouter Castryck and Frederik Vercauteren,
Proceedings of EUROCRYPT 2016.
slides video
PhD Thesis
Optimisations of fully homomorphic encryption
supervised by Prof. dr. ir. Bart Preneel and Prof. dr. ir. Frederik Vercauteren,
KU Leuven, May 2019.