Also known as LWE
problem in machine learning that is conjectured to be hard to solve. Introduced by Oded Regev in 2005, it is a generalization of the parity learning problem
Discovered by embedding cosine similarity (sentence-transformers MiniLM, 384-dim).