# Early Exit Strategies for Approximate k-NN Search in Dense Retrieval

arXiv: https://arxiv.org/abs/2408.04981 · HTML: https://mlsanity.com/paper/2408.04981/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 9 Aug 2024.

## Built on by (most cited first, 4 of 4)

- [Accelerating Retrieval-Augmented Generation](https://mlsanity.com/paper/2412.15246/) (2024)
- [Less Context, Same Performance: A RAG Framework for Resource-Efficient LLM-Based Clinical NLP](https://mlsanity.com/paper/2505.20320/) (2025)
- [GaiaFlow: Semantic-Guided Diffusion Tuning for Carbon-Frugal Search](https://mlsanity.com/paper/2602.15423/) (2026)
- [HNSW with Accuracy Guarantees Using Graph Spanners](https://mlsanity.com/paper/2607.02338/) (2026)

## Abstract (arXiv)

Learned dense representations are a popular family of techniques for encoding queries and documents using high-dimensional embeddings, which enable retrieval by performing approximate k nearest-neighbors search (A-kNN). A popular technique for making A-kNN search efficient is based on a two-level index, where the embeddings of documents are clustered offline and, at query processing, a fixed number N of clusters closest to the query is visited exhaustively to compute the result set. In this paper, we build upon state-of-the-art for early exit A-kNN and propose an unsupervised method based on the notion of patience, which can reach competitive effectiveness with large efficiency gains. Moreover, we discuss a cascade approach where we first identify queries that find their nearest neighbor within the closest t << N clusters, and then we decide how many more to visit based on our patience approach or other state-of-the-art strategies. Reproducible experiments employing state-of-the-art dense retrieval models and publicly available resources show that our techniques improve the A-kNN efficiency with up to 5x speedups while achieving negligible effectiveness losses. All the code used is available at https://github.com/francescobusolin/faiss_pEE