Injecting a few malicious vectors near the centroid exploits centrality-driven hubness in high-dimensional embeddings, causing them to dominate top-k retrievals in up to 99.85% of cases.
Learned Query Optimizer: What is New and What is Next
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
RELOAD achieves up to 2.4x higher robustness and 3.1x greater efficiency than prior RL-based query optimizers on Join Order Benchmark, TPC-DS, and Star Schema Benchmark.
citing papers explorer
-
Can You Trust the Vectors in Your Vector Database? Black-Hole Attack from Embedding Space Defects
Injecting a few malicious vectors near the centroid exploits centrality-driven hubness in high-dimensional embeddings, causing them to dominate top-k retrievals in up to 99.85% of cases.
-
RELOAD: A Robust and Efficient Learned Query Optimizer for Database Systems
RELOAD achieves up to 2.4x higher robustness and 3.1x greater efficiency than prior RL-based query optimizers on Join Order Benchmark, TPC-DS, and Star Schema Benchmark.