Anchor Divergences: contextual semantic similarity for vector search without changing embeddings
victorveitch · x · 2026-10-20
New research from Victor Veitch and collaborators introduces Anchor Divergences. Encoding semantic similarity into vector similarity is a basic building block of AI systems, but fixed notions like cosine similarity can't capture that semantic similarity is fundamentally contextual.
- The method uses distributions over text anchors to change how images are compared — without modifying their embeddings.
- The information-geometric lens both clarifies the apparent contradiction and yields a simple, effective technique for contextual semantics in vector search.
More from Research
- DeepMind's Gemini agent discovers and Lean-verifies new acceleration rules for 11 BB(6) Turing machines — omarsar0 · 2026-10-20
- Atomwell debuts: 9B foundation model shows clean scaling laws on 80M experimental measurements — chaitjo · 2026-10-20
- SFT vs RLVR: how fine-tuning paradigms shape LLM confidence calibration in high-stakes use — burkov · 2026-10-20
- OpenAI dropped 722 AI-generated math papers in a day; can humans read them? — songyoupeng · 2026-10-20
- Skeptic points to a critical paper everyone should read on fly brain simulations — ShahabBakht · 2026-10-20
- UCSB team wins $1M NSF grant to model higher-order interactions in physical systems with AI — ninamiolane · 2026-10-20