Talk:Kyber

Properties

The system is based on module learning with errors (M-LWE) from the field of machine learning [...]

Has this really todo anything with machine learning? I thought LWE is pure lattice theory. 2A01:C22:6E55:5600:D0B8:EB85:18DB:CF1C (talk) 06:08, 29 March 2023 (UTC)[reply]

Some historical connection definitely exists (cf. the title of Shlomi Dolev; Sachin Lodha, eds. (14 June 2017). Cyber Security Cryptography and Machine Learning: First International Conference, CSCML 2017, Beer-Sheva, Israel, June 29-30, 2017, Proceedings. Springer. ISBN 978-3-319-60080-2. OCLC 1001512703.), but it is quite tenuous. I think that removing the words "machine learning" will make the text less confusing. Dimawik (talk) 18:41, 29 March 2023 (UTC)[reply]

The formal name is usually given as "CRYSTALS-Kyber" both to distinguish from the Star Wars crystals as well as to give a clear connection to the digital signature scheme made by the same group: CRYSTALS-Dilithium. This is observed on all of the official NIST submission pages (https://csrc.nist.gov/Projects/post-quantum-cryptography/post-quantum-cryptography-standardization/round-1-submissions, https://csrc.nist.gov/Projects/post-quantum-cryptography/post-quantum-cryptography-standardization/round-3-submissions) as well as many academic publications on the subject (https://eprint.iacr.org/2024/260.pdf, https://eprint.iacr.org/2024/135.pdf, https://eprint.iacr.org/2024/095.pdf). I think the title of the article should be given as "CRYSTALS-Kyber", though the scheme could still be called "Kyber" throughout the rest of the article as a clear and commonly used shorthand. KJack115 (talk) 17:42, 5 March 2024 (UTC)[reply]