{"id":"arxiv/2601.13744","url":"https://arxiv.org/abs/2601.13744","tags":["math.ST"],"score":0,"title":"A Note on k-NN Gating in RAG","domain":"math.ST","source":"arxiv","authors":["Gérard Biau","Claire Boyer"],"reviews":0,"subject":"math.ST","summary":"We propose a statistical proxy framework for retrieval-augmented generation (RAG) that formalizes how language models balance internal predictions with retrieved evidence. We derive an optimal query-level gate, analyze hallucination via retrieval discordance, model query-memory mismatch, and validate the framework numerically on synthetic and real data.","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-01-20","reactions":{}}