Filtered Approximate Nearest Neighbor Search Cost Estimation

2026-02-06

AI Summary

This work addresses the highly variable search cost in hybrid queries arising from the coupling of vector similarity search and structured attribute filtering, which poses significant challenges for effective optimization. The paper proposes E2E, an end-to-end cost estimation framework that, for the first time, explicitly models the correlation between query vector distributions and attribute selectivity. This approach substantially improves the accuracy of cost prediction for filtered approximate nearest neighbor search. Leveraging this estimator, E2E enables an efficient early-termination strategy that achieves 2–3× retrieval speedup on real-world datasets while maintaining high recall. By integrating a deep learning–based cost model with vector indexing and structured filtering mechanisms, E2E establishes a novel paradigm for hybrid query optimization.

Technology Category

• Search and Optimization: Learning to Search
• Data Mining & Knowledge Management: Intelligent Query Processing
• Constraint Satisfaction and Optimization: Search

Application Category

• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Graph Algorithms and Modeling for the Web: Querying, indexing, and retrieval in Web-related graphs
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Hybrid queries combining high-dimensional vector similarity with structured attribute filtering have garnered significant attention across both academia and industry. A critical instance of this paradigm is filtered Approximate k Nearest Neighbor (AKNN) search, where embeddings (e.g., image or text) are queried alongside constraints such as labels or numerical range. While essential for rich retrieval, optimizing these queries remains challenging due to the highly variable search cost induced by combined filters. In this paper, we propose a novel cost estimation framework, E2E, for filtered AKNN search and demonstrate its utility in downstream optimization tasks, specifically early termination. Unlike existing approaches, our model explicitly captures the correlation between the query vector distribution and attribute-value selectivity, yielding significantly higher estimation accuracy. By leveraging these estimates to refine search termination conditions, we achieve substantial performance gains. Experimental results on real-world datasets demonstrate that our approach improves retrieval efficiency by 2x-3x over state-of-the-art baselines while maintaining high search accuracy.

Problem

• filtered approximate nearest neighbor search
• cost estimation
• hybrid queries
• search optimization
• attribute filtering

Innovation

• filtered approximate nearest neighbor
• cost estimation
• vector-attribute correlation
• early termination
• hybrid query optimization