Efficient Filtered-ANN via Learning-based Query Planning

2026-02-20

AI Summary

This work addresses the trade-off between efficiency and recall in filtered approximate nearest neighbor (Filtered ANN) search, where pre-filtering and post-filtering strategies often sacrifice one for the other. To resolve this, the authors propose a learning-based query planning framework that dynamically selects the optimal execution plan at query time. The framework introduces a lightweight learning mechanism that leverages statistical features of both queries and data distributions, combined with a machine learning predictor, to balance filtering effectiveness and ANN search efficiency. It is designed to be backend-agnostic—compatible with any ANN index—and supports diverse filter types. Experimental results demonstrate that the approach achieves up to a 4× speedup over strong baselines while maintaining a recall of at least 90%.

Technology Category

• Search and Optimization: Learning to Search
• Planning, Routing, and Scheduling: Learning for Planning and Scheduling
• Machine Learning: Learning Preferences or Rankings

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Querying, indexing, and retrieval in Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Filtered ANN search is an increasingly important problem in vector retrieval, yet systems face a difficult trade-off due to the execution order: Pre-filtering (filtering first, then ANN over the passing subset) requires expensive per-predicate index construction, while post-filtering (ANN first, then filtering candidates) may waste computation and lose recall under low selectivity due to insufficient candidates after filtering. We introduce a learning-based query planning framework that dynamically selects the most effective execution plan for each query, using lightweight predictions derived from dataset and query statistics (e.g., dimensionality, corpus size, distribution features, and predicate statistics). The framework supports diverse filter types, including categorical/keyword and range predicates, and is generic to use any backend ANN index. Experiments show that our method achieves up to 4x acceleration with>= 90% recall comparing to the strong baselines.

Problem

• Filtered ANN
• vector retrieval
• query planning
• pre-filtering
• post-filtering

Innovation

• Filtered ANN
• learning-based query planning
• pre-filtering vs post-filtering
• vector retrieval
• dynamic execution plan