iFVS: Towards Instance-Optimized Filtered Vector Search

Abdullah Al-Mamun; Walid G. Aref; Yeasir Rayhan

REVIEW 4 major objections 5 minor 30 references

iFVS: Towards Instance-Optimized Filtered Vector Search

T0 review · 4 major / 5 minor · reviewed 2026-08-01 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read iFVS claims that adapting the PQ codebook to each query-filter pair improves the QPS-recall tradeoff of compressed filtered vector search without extra per-vector storage.

desk verdict Novel idea for query-conditioned PQ codebooks in filtered vector search, but the headline recall gain is on training queries; held-out test recall is below the fixed-codebook baseline, so the core claim doesn't hold. read the letter →

arxiv 2607.22922 v1 pith:BADEGEXF submitted 2026-07-24 cs.DB

Abdullah Al-Mamun , Yeasir Rayhan , Walid G. Aref This is my paper

classification cs.DB

keywords filteredvectorsearchproductquantizationinstance-optimizedcodebookadaptationquery-awarePQapproximatenearestneighborQPS-recalltradeoff

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Filtered vector search combines similarity ranking with relational predicates, and quantized indexes like PQ lose recall because a single fixed codebook is blind to how the filter reshapes the relevant vector space. iFVS proposes to generate a query- and filter-specific perturbation of the base PQ codebook, plus a predicate-derived dimension weighting, so distance estimates are conditioned on the actual query-predicate pair. Starting from the standard PQ codebook with zero adjustment, it trains the perturbation through hard-negative mining and an anchor regularizer, keeping the stored PQ codes and base codebook fixed. On SIFT1M and SIFT10M workloads, iFVS reports consistently better Recall@100 at higher or comparable QPS than fixed-codebook PQ baselines across 1%, 5%, and 10% selectivity bins, with compact index sizes as a byproduct. If correct, this shows that quantization for filtered search can be instance-optimized rather than workload-agnostic.

What carries the argument

The central mechanism is the query-specific codebook perturbation Delta = alpha * W, where alpha is produced by hashing the query-predicate pair into a shared memory bank M (h independent hash functions summed over B rows) and W is a learned tensor of shape M x r x K x d; the effective codebook is C_iFVS = C_PQ + Delta, with the predicate also encoded into a dimension-weight vector w_f = softplus(w_z) via a lightweight Transformer. This construction turns a single static PQ codebook into a family of filter-aware codebooks, and because alpha is initialized to zero and w_f to one, training starts from the exact PQ baseline. The mechanism's job is to condition distance estimation on the specifi

What would settle it

Run the same iFVS training pipeline on a published filtered-search benchmark with real metadata predicates and compare Recall@100 vs QPS at 1% selectivity: if iFVS no longer exceeds Pre_PQ_ADC on both recall and QPS, or if its test recall with a large memory bank falls below the fixed-codebook baseline, the instance-optimization claim would fail on that ground.

Watch

Extended reading notes

Core claim

The paper's central claim is that the precision penalty of product quantization in filtered vector search can be partially recovered by making the codebook itself a function of the query and the filter predicate. Concretely, iFVS fingerprints each query-predicate pair into a shared memory bank, sums the hash lookups to form an adjustment vector alpha, and combines alpha with a learned tensor W to perturb the base PQ codebook; a lightweight Transformer encodes the predicate into a weight vector that reweights dimensions during scoring. The perturbation is initialized to zero and the weights to one, so the model degenerates to plain PQ before training; training then adapts alpha and w_f using

Load-bearing premise

The load-bearing premise is that the synthetic filtered workload—predicates imposed on one to three fixed SIFT dimensions with selectivities 1%, 5%, 10% and exact ground truth over predicate-passing vectors—faithfully represents real filtered vector search workloads; if real predicates interact with the vector space differently, the learned codebook adjustments and weight reweighting may not transfer.

Editorial extensions

If this is right

On SIFT10M at 1% selectivity, iFVS reports Recall@100 of 0.774 at 334 QPS versus 0.682 at 114 QPS for Pre_PQ_ADC, a direct demonstration that codebook adaptation improves the QPS-recall frontier.
iFVS dominates post-filtering baselines on both datasets and all selectivity bins, and beats all PQ-compressed in-filtering baselines; only raw-vector in-filtering (IVFFlat, HNSWFlat) can exceed its recall at larger selectivities, at lower QPS.
Scaling from 1M to 10M vectors, iFVS's QPS drops only 1.33x at 1% selectivity, while exact pre-filtering slows by 20.2x and Pre_PQ_ADC by 13.6x, suggesting favorable scaling for large quantized FVS.
The compact index size (89.9 MB on SIFT1M, 220.5 MB on SIFT10M) comes at an offline construction cost of 57 and 89 minutes, respectively; this is the primary practical tradeoff.
Memory-bank size M controls a specialization-generalization tradeoff: larger M raises training recall (0.909 to 0.971 on SIFT1M) but lowers test recall (0.749 to 0.702), so generalization to unseen queries is sensitive to this hyperparameter.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

If real FVS workloads distribute filters over many attribute types rather than a few SIFT dimensions, iFVS's learned adjustments may not transfer; the paper's own future-work list of recently proposed filtered-query benchmarks makes this the open question.
The perturbation mechanism is codebook-level, so the same instance-optimization idea could be applied to other quantization schemes (e.g., OPQ, additive quantization) or even to graph traversal scores, not just PQ.
The observed test-recall decline with larger memory banks suggests overfitting to training query-predicate pairs; adding regularization or a small meta-learned prior on alpha could improve generalization to filters outside the training distribution.
The predicate-derived dimension weighting w_f is decoupled from the codebook perturbation and could be tested in full-precision filtered search, where it might provide a cheap dynamic feature-weighting baseline independent of quantization.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

You should know one thing going in: the paper's big reported numbers are averages over training and test queries, and when you separate them the test recall is below the fixed-codebook baseline. Section 3.3.1 says the figures report the average; Table 2 gives the split. At 1% selectivity on SIFT10M, the 0.774 Recall@100 matches memory bank M=45,000, average of train 0.8906 and test 0.6572. Pre_PQ_ADC gets 0.682. So on held-out queries, iFVS underperforms the fixed-codebook PQ it claims to beat. The authors' own future-work item (b) "improve generalization over unseen queries" confirms they know this. That's not a small caveat; it undercuts the central claim of the paper.

What's actually new and worth credit: the mechanism — perturbing a fixed PQ codebook with a query-filter-specific adjustment learned via a memory bank, plus dimension reweighting based on the predicate — is a genuinely new combination for FVS. The training-side gains are real, and the QPS-recall tradeoff on training queries is strong. The paper also honestly surveys a wide set of baselines, even if the comparison is apples-to-oranges: iFVS is a learned, workload-tuned method, and the baselines are not.

The evaluation has other soft spots. No code, no hyperparameters, no precise training objective. The workload is synthetic predicates on a few SIFT dimensions, so we don't know if the learned adjustments transfer to realistic filter predicates. And the method is never compared against the query-aware quantization baselines it cites (OOD-DiskANN, query-aware quantization for MIPS), so the positioning is incomplete. These are fixable in principle; the generalization gap is not — it's a property of the method as presented, not just the write-up.

For a workshop paper, the idea is worth one serious look. If I were the editor, I'd send it to a referee who knows FVS, with explicit instructions to check the train/test reporting. But in its current form, the central claim is unsupported, and I would not cite it as evidence that query-specific codebook adaptation improves ranking for unseen queries.

Referee Report

4 major / 5 minor

Summary. The paper presents iFVS, a learned method for filtered vector search over product-quantized databases. For each query-predicate pair, iFVS fingerprints the pair into a shared memory bank, uses the resulting coefficients to perturb the original PQ codebook, and reweights query and database dimensions with a filter-aware weight vector; the stored PQ codes remain fixed. The method is evaluated on synthetic filtered workloads over SIFT1M and SIFT10M with selectivity 0.01, 0.05, and 0.10, against 13 pre-, post-, and in-filtering baselines. The paper reports QPS-recall tradeoffs and claims that iFVS improves ranking quality over fixed-codebook PQ baselines, e.g., Recall@100 0.774 at 334 QPS vs. Pre_PQ_ADC's 0.682 at 114 QPS on SIFT10M at 1% selectivity.

Significance. If validated, the central idea is relevant and timely: instance-optimizing the quantizer to a filtered workload is a natural extension of the instance-optimized database agenda, and the paper correctly notes that prior query-aware quantization has not been studied for FVS. The evaluation design has strengths: many baselines, separate training/test recall reporting in Table 2, and an anchor regularizer that limits drift from the original PQ codebook. However, the current evidence does not establish the main claim for unseen queries, because the headline comparisons use the average of training and test recall; the held-out test recall is below the corresponding PQ baseline. The absence of training details and the reliance on one synthetic workload further limit the contribution. The direction is plausible, but the paper needs a substantial revision of the evaluation and claims.

major comments (4)

[§3.3.1–3.3.2, Table 2] The headline comparison uses the train/test average, not held-out recall. The reported SIFT10M number 0.774 Recall@100 is the average of train recall 0.8906 and test recall 0.6572 at M=45,000 (Table 2); the SIFT1M number 0.836 is similarly the average at M=45,000. The best held-out test recall in Table 2 is 0.6706 (SIFT10M, M=4,096), which is below Pre_PQ_ADC's 0.682 reported in §3.3.2. Thus, on unseen queries, iFVS's ranking is worse than this fixed-codebook PQ baseline, contradicting the central claim that query-specific codebook adaptation improves ranking for a representative workload. Section 4(b), which lists improving generalization over unseen queries as future work, confirms that the current method does not deliver this.
[§2, Learning C_iFVS and w_f] The training objective is not specified. No loss function, hard-negative mining details, number of hard negatives, optimizer, learning rate, batch size, number of epochs, or values for hyperparameters r and h are given. The anchor regularizer is mentioned but its weight is not specified. Since iFVS is a learned component and the paper's evidence is purely empirical, this omission is load-bearing for reproducibility and for understanding whether the reported gains come from the architecture or from a particular training schedule.
[§3.1, §3.3.1] The 'representative workload' is entirely synthetic: predicates are imposed on one to three fixed SIFT dimensions, with selectivity bins [0.01, 0.05, 0.10]. The paper does not report which dimensions, how many predicates, or how predicate difficulty varies. Because iFVS is explicitly instance-optimized, every conclusion is tied to this single synthetic workload. The conclusion correctly identifies recent filtered-query benchmarks as future work, but the abstract and introduction state the improvement without this caveat. The evaluation should either include a second workload or the claims should be re-scoped to the tested setting.
[§3.3.1, Figures 1–2] The iFVS entries are single operating points, whereas all baselines are plotted as tradeoff curves by sweeping ef/nprobe. A single point that lies above a baseline curve does not by itself demonstrate a superior QPS-recall tradeoff. To support the 'tradeoff' language, iFVS should be evaluated at multiple operating points (e.g., by varying the number of candidates scored or the memory-bank size) and plotted as a curve. If only one point is available, the paper should make dominance explicit rather than claiming an improved tradeoff.

minor comments (5)

[§3.3.1–3.3.2] The sentence 'The figures report the average recall across both sets' should be prominently repeated before every headline recall number; otherwise readers naturally interpret 0.774 and 0.836 as test-set results.
[§2] Notation is inconsistent: C_eff appears once, while C_iFVS is used elsewhere; w_z and w_f are used interchangeably. Define all symbols in one place and use them consistently.
[Tables 1 and 2] The symbol M denotes HNSW graph connections in Table 1 and memory-bank size in Table 2. This is confusing; rename one of them.
[Figures 1–2] The captions do not state which memory-bank size M is used for iFVS. The reader must infer it from the averages in Table 2. Add this information to the caption.
[§3.2, §3.3.7] There are a few typos/inconsistencies: 'calucation' in §3.2, and 'Pre_FlatL2' in §3.3.7 should be 'Pre_IndexFlatL2'.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: iFVS is a learned empirical method; the headline train/test averaging is an evaluation caveat, not a definitional reduction.

full rationale

The paper contains no derivation chain in which a predicted quantity is identical by construction to a fitted input. iFVS is a supervised, learned perturbation of a fixed PQ codebook: training uses ground-truth positives and hard negatives (Section 2), and evaluation uses held-out test queries (Section 3.3.1). Citations to instance-optimization work ([3], [5], [6]) are motivational, not load-bearing; no uniqueness theorem or ansatz is imported from the authors' prior work. The only notable concern is evaluative: Section 3.3.1 states that figures report average recall across training and test queries, and Table 2 reports separately; Section 3.3.2's headline comparison uses that average, while Table 2 shows SIFT10M test recall between 0.6572 and 0.6706, below the Pre_PQ_ADC recall of 0.682. This is an overfitting or reporting caveat, acknowledged in Section 3.3.5 and future work (b), but because the test-split numbers are explicitly disclosed, the result is not a disguised pure prediction nor equivalent to its input by definition. Hence a low score of 2 rather than a circularity flag.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

No physics-like entities are invented. The method introduces learned parameters (memory bank M, adjustment tensor W, filter Transformer), which are internal model weights, not separate scientific entities. The main ledger items are the unspecified hyperparameters and the unstated workload design choices.

free parameters (5)

Memory bank size M = Swept: 4096, 8192, 16384, 32768, 45000 (Table 2)
Controls the capacity of query-filter fingerprints; the paper shows it trades off training recall versus test recall, but does not justify the chosen default. The comparison in Figures 1 and 2 presumably uses one of these values.
Adjustment directions r = Not reported
Hyperparameter specifying the number of directions the codebook adjustment can take; never specified in the evaluation.
Number of hash functions h = Not reported
Fingerprinting the query-filter pair requires h hash functions; never specified.
Training hyperparameters = Not reported
Learning rate, number of epochs, hard-negative mining parameters, anchor regularizer strength are all unstated; these directly affect the reported recall/QPS scores.
Filter workload dimensions = One to three fixed SIFT dimensions, unspecified which
The synthetic predicates are sampled over 'one to three of these fixed dimensions'; the specific dimensions are not stated, so the workload could be easier or harder depending on which spatial cells are chosen.

assumptions (3)

domain assumption SIFT1M/SIFT10M and their 10K query vectors are a representative base for FVS evaluation.
Section 3.1 uses only SIFT datasets with synthetic filter predicates; no real filtered workloads or transformer-based embeddings are tested.
domain assumption A query-specific codebook perturbation with only additive updates (Delta = alpha * W) preserves ranking quality across the whole candidate pool.
The method is trained on top-100 ground truth with hard negatives; the paper does not prove or ablate that the additive perturbation generalizes to all candidate rankings, only that it improves recall@100.
domain assumption Standard PQ codebook behavior (ADC-style scoring) generalizes when the codebook is perturbed.
The score formula in Equation 3 uses the perturbed codebook entries as if they are ordinary PQ centroids; if w_f reweighting changes the distance space, the nearest-neighbor semantics are harder to interpret.

how reviews work

0 comments

Cite this review

Pith. "Pith review of iFVS: Towards Instance-Optimized Filtered Vector Search." pith.science (2026). https://pith.science/paper/BADEGEXF

@misc{pith2026260722922,
  author       = {Pith},
  title        = {Pith review of: iFVS: Towards Instance-Optimized Filtered Vector Search},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BADEGEXF}},
  note         = {Machine review of arXiv:2607.22922}
}

read the original abstract

Filtered vector search (FVS) is increasingly important in modern AI + DB systems, where vector similarity search is combined with relational predicates. Quantization plays a vital role in these systems by enabling query processing over large vector datasets. However, lossy approaches, e.g., Product Quantization (PQ), incur a precision penalty during distance calculation, thereby negatively impacting the query recall performance. This problem becomes more challenging in FVS because the relevant vector space can change with the relational predicate and selectivity. Motivated by the success of instance-optimized database system components, we introduce iFVS, an Instance-Optimized Filtered Vector Search technique. Given a fixed, quantized vector dataset, and a representative workload of filtered vector queries, iFVS adopts a query-specific codebook generation approach for FVS that is instance-optimized towards a certain dataset and query workload. Instead of using a fixed codebook for all queries, iFVS conditions distance estimation on both the query vector and the filter predicate. This enables more accurate ranking over compressed vectors while preserving compact per-vector storage. Experiments show that iFVS improves the Queries Per Second (QPS)-recall tradeoff across several filter selectivity bins compared with fixed-codebook quantized FVS baselines.

Figures

Figures reproduced from arXiv: 2607.22922 by the authors.

**Figure 1.** FVS over the SIFT1M dataset. 0.0 0.2 0.4 0.6 0.8 1.0 Recall@k=100 (Selectivity 0.01) 10 1 10 2 QPS (queries/sec) 0.0 0.2 0.4 0.6 0.8 1.0 Recall@k=100 (Selectivity 0.05) 10 1 10 2 0.2 0.4 0.6 0.8 1.0 Recall@k=100 (Selectivity 0.10) 10 0 10 1 10 2 Pre_IndexFlatL2 Pre_IndexFlatL2_PQ Pre_IndexPQ_ADC Pre_IndexPQ_SDC Pre_SDC_numpy Post_HNSWPQ Post_IVFPQ In_HNSWPQ In_IVFPQ Post_HNSWFlat Post_IVFFlat In_HNSWFlat In_IVFFlat … view at source ↗

**Figure 2.** FVS over the SIFT10M dataset. HNSWPQ IVFPQ HNSWFlat IVFFlat iFVS Index Type 0 200 400 600 Size (MB) Data Size Index Size (a) SIFT 1M. HNSWPQ IVFPQ HNSWFlat IVFFlat iFVS Index Type 0 2500 5000 7500 Size (MB) Data Size Index Size (b) SIFT 10M [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** Index size comparison. HNSWPQ IVFPQ HNSWFlat IVFFlat iFVS Index Type 10 1 10 2 10 3 Time (seconds) (a) SIFT 1M. HNSWPQ IVFPQ HNSWFlat IVFFlat iFVS Index Type 10 3 10 4 Time (seconds) (b) SIFT 10M [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: Index construction time. 3.3.7 Scaling from SIFT1M to SIFT10M. iFVS scales favorably as the dataset grows by 10×. At 1% selectivity, QPS decreases from 442.6 on SIFT1M to 333.8 on SIFT10M, a slowdown of only 1.33×. In contrast, Pre_FlatL2 slows down by 20.2×, and Pre_P…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

30 extracted references · 1 linked inside Pith

[1]

SIFT Dataset

[n.d.]. SIFT Dataset. http://corpus-texmex.irisa.fr/. Dataset webpage. Accessed: 2026

2026
[2]

Anas Ait Aomar, Karima Echihabi, Marco Arnaboldi, Ioannis Alagiannis, Damien Hilloulin, and Manal Cherkaoui. 2025. RWalks: Random Walks as Attribute Diffusers for Filtered Vector Search.Proceedings of the ACM on Management of Data3, 3 (2025), 1–26

2025
[3]

Abdullah Al-Mamun, Hao Wu, Qiyang He, Jianguo Wang, and Walid G Aref
[4]

Yannis Chronis, Helena Caminal, Yannis Papakonstantinou, Fatma Özcan, and Anastasia Ailamaki. 2025. Filtered Vector Search: State-of-the-Art and Research Opportunities.Proceedings of the VLDB Endowment18, 12 (2025), 5488–5492

2025
[5]

Jialin Ding, Ryan Marcus, Andreas Kipf, Vikram Nathan, Aniruddha Nrusimha, Kapil Vaidya, Alexander van Renen, and Tim Kraska. 2022. Sagedb: An instance- optimized data analytics system.Proceedings of the VLDB Endowment15, 13 (2022)

2022
[6]

Jialin Ding, Umar Farooq Minhas, Badrish Chandramouli, Chi Wang, Yinan Li, Ying Li, Donald Kossmann, Johannes Gehrke, and Tim Kraska. 2021. Instance- optimized data layouts for cloud analytics workloads. InProceedings of the 2021 International Conference on Management of Data. 418–431

2021
[7]

Matthijs Douze, Alexandr Guzhva, Chengqi Deng, Jeff Johnson, Gergely Szilvasy, Pierre-Emmanuel Mazaré, Maria Lomeli, Lucas Hosseini, and Hervé Jégou. 2025. The faiss library.IEEE Transactions on Big Data(2025)

2025
[8]

Siddharth Gollapudi, Neel Karia, Varun Sivashankar, Ravishankar Krishnaswamy, Nikit Begwani, Swapnil Raz, Yiyong Lin, Yin Zhang, Neelam Mahapatro, Premku- mar Srinivasan, et al. 2023. Filtered-diskann: Graph algorithms for approximate nearest neighbor search with filters. InProceedings of the ACM Web Conference

2023

Show all 30 references

[9]

Gray and David L

Robert M. Gray and David L. Neuhoff. 2002. Quantization.IEEE transactions on information theory44, 6 (2002), 2325–2383

2002
[10]

Patrick Iff, Paul Brügger, Marcin Chrapek, Maciej Besta, and Torsten Hoefler. 2025. Benchmarking Filtered Approximate Nearest Neighbor Search Algorithms on Transformer-based Embedding Vectors.arXiv preprint arXiv:2507.21989(2025)

2025
[11]

Shikhar Jaiswal, Ravishankar Krishnaswamy, Ankit Garg, Harsha Vardhan Simhadri, and Sheshansh Agrawal. 2022. Ood-diskann: Efficient and scalable graph anns for out-of-distribution queries.arXiv preprint arXiv:2211.12850(2022)

2022 arXiv
[12]

Herve Jegou, Matthijs Douze, and Cordelia Schmid. 2010. Product quantization for nearest neighbor search.IEEE transactions on pattern analysis and machine intelligence33, 1 (2010), 117–128

2010
[13]

Tim Kraska, Alex Beutel, Ed H Chi, Jeffrey Dean, and Neoklis Polyzotis. 2018. The case for learned index structures. InProceedings of the 2018 international conference on management of data. 489–504

2018
[14]

Zhaoheng Li, Silu Huang, Wei Ding, Yongjoo Park, and Jianjun Chen. 2025. SIEVE: Effective Filtered Vector Search with Collection of Indexes.Proc. VLDB Endow.18, 11 (July 2025), 4723–4736

2025
[15]

Anqi Liang, Pengcheng Zhang, Bin Yao, Zhongpu Chen, Yitong Song, and Guangxu Cheng. 2024. UNIFY: Unified Index for Range Filtered Approximate Nearest Neighbors Search.Proc. VLDB Endow.18, 4 (2024), 1118–1130

2024
[16]

Jiayi Liu, Yunan Zhang, Chenzhe Jin, Aditya Gupta, Shige Liu, and Jianguo Wang
[17]

Duo Lu, Helena Caminal, Manos Chatzakis, Yannis Papakonstantinou, Yannis Chronis, Vaibhav Jain, and Fatma Özcan. 2026. An In-Depth Study of Filter- Agnostic Vector Search on a PostgreSQL Database System:[Experiments & Anal- ysis].Proceedings of the ACM on Management of Data4, ...

2026
[18]

Yu A Malkov and Dmitry A Yashunin. 2018. Efficient and robust approximate nearest neighbor search using hierarchical navigable small world graphs.IEEE transactions on pattern analysis and machine intelligence42, 4 (2018), 824–836

2018
[19]

Jason Mohoney, Anil Pacaci, Shihabur Rahman Chowdhury, Ali Mousavi, Ihab F Ilyas, Umar Farooq Minhas, Jeffrey Pound, and Theodoros Rekatsinas. 2023. High-throughput vector similarity search in knowledge graphs.Proceedings of the ACM on Management of Data1, 2 (2023), 1–25

2023
[20]

James Jie Pan, Jianguo Wang, and Guoliang Li. 2024. Survey of vector database management systems.The VLDB Journal33, 5 (2024), 1591–1615

2024
[21]

Liana Patel, Peter Kraft, Carlos Guestrin, and Matei Zaharia. 2024. Acorn: Per- formant and predicate-agnostic search over vector embeddings and structured data.Proceedings of the ACM on Management of Data2, 3 (2024), 1–27

2024
[22]

pgvector. [n.d.]. pgvector: Open-source vector similarity search for Postgres. https://github.com/pgvector/pgvector. GitHub repository. Accessed: 2026

2026
[23]

Junjie Song, Yu Liu, Guoyu Hu, Zhongle Xie, Ming Yang, Beng Chin Ooi, and Ke Zhou. 2026. FAVOR: Efficient Filter-Agnostic Vector ANNS Based on Selectivity- Aware Exclusion Distances.Proceedings of the ACM on Management of Data4, 3 (SIGMOD (2026), 1–25

2026
[24]

Mihail Stoian, Johannes Thürauf, Andreas Zimmerer, Alexander van Renen, and Andreas Kipf. 2025. Instance-Optimized String Fingerprints (Extended Abstracts). Applied AI for Database Systems and Applications (AIDB) at VLDB(2025)

2025
[25]

Jianguo Wang, Xiaomeng Yi, Rentong Guo, Hai Jin, Peng Xu, Shengjun Li, Xi- angyu Wang, Xiangzhou Guo, Chengming Li, Xiaohai Xu, et al. 2021. Milvus: A purpose-built vector data management system. InProceedings of the 2021 international conference on management of data. 2614–2627

2021
[26]

Jin Zhang, Defu Lian, Haodi Zhang, Baoyun Wang, and Enhong Chen. 2023. Query-aware quantization for maximum inner product search. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 37. 4875–4883

2023
[27]

Xiang Zhang, Chao Zhang, Ju Fan, Guoliang Li, and Xiaoyong Du. 2026. VecBench: A Controllable Benchmark for Filtered Vector Search:[Experiments & Analysis].Proceedings of the ACM on Management of Data4, 3 (SIGMOD (2026), 1–27

2026
[28]

Chaoji Zuo, Miao Qiao, Wenchao Zhou, Feifei Li, and Dong Deng. 2024. Serf: Seg- ment graph for range-filtering approximate nearest neighbor search.Proceedings of the ACM on Management of Data2, 1 (2024), 1–26. 5

2024
[2025]

Surveys58, 4 (2025), 1–37

A survey of learned indexes for the multi-dimensional space.Comput. Surveys58, 4 (2025), 1–37. 4

2025
[2026]

Fast Vector Search in PostgreSQL: A Decoupled Approach.. InCIDR

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.

[1] [1]

SIFT Dataset

[n.d.]. SIFT Dataset. http://corpus-texmex.irisa.fr/. Dataset webpage. Accessed: 2026

2026

[2] [2]

Anas Ait Aomar, Karima Echihabi, Marco Arnaboldi, Ioannis Alagiannis, Damien Hilloulin, and Manal Cherkaoui. 2025. RWalks: Random Walks as Attribute Diffusers for Filtered Vector Search.Proceedings of the ACM on Management of Data3, 3 (2025), 1–26

2025

[3] [3]

Abdullah Al-Mamun, Hao Wu, Qiyang He, Jianguo Wang, and Walid G Aref

[4] [4]

Yannis Chronis, Helena Caminal, Yannis Papakonstantinou, Fatma Özcan, and Anastasia Ailamaki. 2025. Filtered Vector Search: State-of-the-Art and Research Opportunities.Proceedings of the VLDB Endowment18, 12 (2025), 5488–5492

2025

[5] [5]

Jialin Ding, Ryan Marcus, Andreas Kipf, Vikram Nathan, Aniruddha Nrusimha, Kapil Vaidya, Alexander van Renen, and Tim Kraska. 2022. Sagedb: An instance- optimized data analytics system.Proceedings of the VLDB Endowment15, 13 (2022)

2022

[6] [6]

Jialin Ding, Umar Farooq Minhas, Badrish Chandramouli, Chi Wang, Yinan Li, Ying Li, Donald Kossmann, Johannes Gehrke, and Tim Kraska. 2021. Instance- optimized data layouts for cloud analytics workloads. InProceedings of the 2021 International Conference on Management of Data. 418–431

2021

[7] [7]

Matthijs Douze, Alexandr Guzhva, Chengqi Deng, Jeff Johnson, Gergely Szilvasy, Pierre-Emmanuel Mazaré, Maria Lomeli, Lucas Hosseini, and Hervé Jégou. 2025. The faiss library.IEEE Transactions on Big Data(2025)

2025

[8] [8]

Siddharth Gollapudi, Neel Karia, Varun Sivashankar, Ravishankar Krishnaswamy, Nikit Begwani, Swapnil Raz, Yiyong Lin, Yin Zhang, Neelam Mahapatro, Premku- mar Srinivasan, et al. 2023. Filtered-diskann: Graph algorithms for approximate nearest neighbor search with filters. InProceedings of the ACM Web Conference

2023