arxiv: 2605.03581 · v1 · submitted 2026-05-05 · 💻 cs.CR

Recognition: unknown

ZK-Value: A Practical Zero-Knowledge System for Verifiable Data Valuation

Zhaoyu Wang (1) , Pingchuan Ma (2) , Zhantong Xue (1) , Yuguang Zhou (1) , Qixin Zhang (3) , Xiaoqin Zhang (2) , Shuai Wang (1) ((1) HKUST , Hong Kong SAR

show 5 more authors

(2) Zhejiang University of Technology Hangzhou China (3) Nanyang Technological University Singapore)

Authors on Pith no claims yet

Pith reviewed 2026-05-07 15:30 UTC · model grok-4.3

classification 💻 cs.CR

keywords zero-knowledge proofsdata valuationShapley valueslocality-sensitive hashingverifiable computationdata marketplaces

0 comments

The pith

ZK-Value makes Shapley data valuations verifiable with zero-knowledge proofs that match exact accuracy and run in practical time.

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

The paper seeks to resolve the tension between privacy and verifiability in data marketplaces by creating a system where Shapley value attributions can be checked by anyone without access to the data. It introduces an approximation based on locality-sensitive hashing to avoid expensive pairwise calculations and a zero-knowledge protocol tailored to prove those approximations efficiently. If the system works as described, marketplaces could operate with transparent and auditable payment allocations while protecting contributor data. This addresses why existing ZK solutions were not usable in practice due to their slow proof times.

Core claim

ZK-Value is a fully co-designed architecture consisting of LSH-Shapley for valuation via bucket collision counts, ZK-LSH-Shapley for proving histogram-encoded counts, and optimizations including super-oracle batching and sparsity skipping. It achieves valuation quality within 0.033 AUROC of exact KNN-Shapley on 12 datasets, with proof generation in seconds to minutes and verification under 4.6 seconds, outperforming other ZK baselines by 12.6x to 68.1x.

What carries the argument

The LSH-Shapley primitive, which approximates Shapley values by counting collisions within hash buckets rather than computing all pairwise distances, and the accompanying ZK-LSH-Shapley protocol that proves these counts using compact bucket-level histograms instead of large per-pair tensors.

If this is right

Market operators can publish verifiable Shapley values for independent checking by any party.
Data providers receive payments based on auditable contributions without exposing their data.
The system enables practical deployment in marketplaces due to fast proof generation and verification.
It substantially reduces proving overhead compared to previous zero-knowledge valuation methods.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

This co-design of approximation and proof could inspire similar optimizations for other expensive ML computations in ZK settings.
Widespread use might boost trust and participation in data marketplaces by ensuring transparent attribution.
The approach may extend to valuing contributions in federated learning scenarios where privacy is also key.

Load-bearing premise

The locality-sensitive hashing approximation must maintain valuation quality close to exact methods across the range of datasets used in data marketplaces.

What would settle it

Testing on an additional dataset where the AUROC gap from exact KNN-Shapley exceeds 0.033 on average or where proof generation times fall outside the reported seconds-to-minutes range would challenge the claims.

Figures

Figures reproduced from arXiv: 2605.03581 by (2) Zhejiang University of Technology, (3) Nanyang Technological University, China, Hangzhou, Hong Kong SAR, Pingchuan Ma (2), Qixin Zhang (3), Shuai Wang (1) ((1) HKUST, Singapore), Xiaoqin Zhang (2), Yuguang Zhou (1), Zhantong Xue (1), Zhaoyu Wang (1).

**Figure 1.** Figure 1: The verifiability gap in a data marketplace. view at source ↗

**Figure 2.** Figure 2: End-to-end architecture of ZK-Value. Four protocol phases. ZK-Value’s valuation pipeline runs in four sequential phases ( view at source ↗

**Figure 3.** Figure 3: ZK-Value scalability along the four workload axes (𝑑, 𝑁 ,𝑇 ,𝐶). Hashing module (85.4%) Layer 1 (histogram consistency) (5.5%) Layer 2 (lookup consistency) (8.8%) Layer 3 (weight + aggregation) (0.2%) Layer 4 (input/output reconstruction) (0.1%) view at source ↗

**Figure 4.** Figure 4: ZK-Value proving-time decomposition. 7 RELATED WORK Privacy-preserving and verifiable data valuation. The Shapley formulation of data valuation has produced a rich algorithmic line [13, 14, 20, 21, 24, 47, 51], all of which assume a trusted valuator with plaintext access to seller data. A second line addresses the resulting privacy tension via differentially private estimators [48] or secure-MPC prototyp… view at source ↗

read the original abstract

Data valuation is a foundational task in data marketplaces, where a Shapley-value attribution determines how a buyer's payment is distributed among data providers. Typically, the marketplace operator runs this attribution alone, requiring participants and external auditors to trust scores they cannot independently recompute on the underlying private data. While zero-knowledge proofs (ZKPs) can theoretically reconcile this conflict between privacy and verifiability, existing ZK valuation systems fail to scale to real-world marketplace demands due to prohibitive proving times or the requirement to disclose validation cohorts. We present ZK-Value, a practical, end-to-end ZK data-valuation system. Our solution bridges the scalability gap through a fully co-designed architecture: (1) LSH-Shapley, a locality-based valuation primitive that replaces expensive pairwise distance metrics with per-bucket collision counts; (2) ZK-LSH-Shapley, a tailored ZKP protocol that drastically reduces witness size by encoding these counts into bucket-level histograms rather than naive per-pair tensors; and (3) structural proof-system optimizations, specifically super-oracle batching and sparsity skipping. Evaluated across 12 standard datasets, ZK-Value delivers valuation quality on par with state-of-the-art baselines (within 0.033 AUROC of exact KNN-Shapley), while generating proofs in seconds to minutes and outperforming specialized ZK baselines by 12.6x to 68.1x in proving time, with verification in under 4.6 s.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

ZK-Value gives a concrete LSH-plus-ZKP pipeline for verifiable Shapley valuation that runs in seconds to minutes on standard data, but the utility claims rest only on observed AUROC gaps without bounds or robustness checks.

read the letter

The paper's core contribution is a co-designed system that swaps exact pairwise distances in KNN-Shapley for LSH bucket collision counts, then builds a tailored ZKP around bucket histograms plus batching and sparsity tricks. This lets them prove the approximate valuation without shipping the full data or the full distance matrix. On the 12 datasets they report, the resulting scores stay within 0.033 AUROC of the exact baseline while cutting proving time by 12x–68x versus prior ZK valuation work and keeping verification under 5 seconds. That combination of numbers is the practical advance worth noting; the architecture choices (histogram encoding, super-oracle batching) are the parts that actually move the needle on performance rather than just applying an off-the-shelf proof system to an existing algorithm. The empirical speed and quality results are the strongest part of the write-up. The soft spot is exactly the one the stress-test flags: the “on par” claim is an observed gap across those 12 sets with no concentration bound on the collision-count estimator, no sweep on hash count or bucket width, and no evaluation on high-intrinsic-dimension or non-uniform data where LSH is known to degrade. Because the ZKP only certifies the approximate protocol, any systematic bias in the LSH step directly undercuts the verifiability guarantee for the intended use case. The paper does not appear to contain internal contradictions or circular reasoning, and the citation pattern looks standard for applied crypto work. This is the kind of system paper that data-marketplace and applied-cryptography readers would want to see, even if they end up needing the theoretical gaps filled in revision. It is coherent enough on its own terms to deserve a serious referee rather than a desk reject.

Referee Report

1 major / 2 minor

Summary. The manuscript presents ZK-Value, a practical end-to-end zero-knowledge system for verifiable data valuation using Shapley values in data marketplaces. It introduces LSH-Shapley, an approximation primitive that replaces pairwise distance computations in KNN-Shapley with per-bucket collision counts from locality-sensitive hashing; ZK-LSH-Shapley, a tailored ZKP protocol that encodes collision counts into bucket-level histograms to reduce witness size; and structural optimizations including super-oracle batching and sparsity skipping. Empirical evaluation across 12 standard datasets claims valuation quality within 0.033 AUROC of exact KNN-Shapley, proving times of seconds to minutes (12.6x–68.1x faster than specialized ZK baselines), and verification times under 4.6 s.

Significance. If the central claims hold, the work would be significant for enabling scalable, privacy-preserving, and auditable data valuation without requiring trust in the marketplace operator or disclosure of validation data. The co-designed LSH approximation and histogram-based ZKP, together with the reported speedups, directly address the scalability limitations of prior ZK valuation systems. The multi-dataset empirical validation provides concrete evidence of practical utility, though additional theoretical grounding on the approximation would further strengthen the contribution.

major comments (1)

[Evaluation] Evaluation section (abstract and reported results on 12 datasets): The headline claim that LSH-Shapley yields valuations within 0.033 AUROC of exact KNN-Shapley rests solely on empirical gaps without concentration bounds on the collision-count estimator, sensitivity analysis for LSH parameters (hash functions, bucket width, number of tables), standard deviations or error bars on the AUROC figures, or targeted evaluation on datasets with high intrinsic dimension or non-uniform densities where LSH locality sensitivity is known to degrade. Because ZK-LSH-Shapley only proves the approximate protocol, any systematic bias in the underlying primitive directly undermines the 'on par' utility guarantee.

minor comments (2)

[Abstract] The abstract states that ZK-Value 'outperforms specialized ZK baselines by 12.6x to 68.1x' but does not name the baselines or provide citations to them.
Notation for LSH parameters (e.g., number of hash tables, bucket width) is introduced without a consolidated symbol table, which would aid readability of the protocol description.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their constructive feedback on our manuscript. The major comment raises important points about strengthening the empirical evaluation of the LSH-Shapley approximation. We address this below and will incorporate targeted improvements in the revision to better support our claims.

read point-by-point responses

Referee: Evaluation section (abstract and reported results on 12 datasets): The headline claim that LSH-Shapley yields valuations within 0.033 AUROC of exact KNN-Shapley rests solely on empirical gaps without concentration bounds on the collision-count estimator, sensitivity analysis for LSH parameters (hash functions, bucket width, number of tables), standard deviations or error bars on the AUROC figures, or targeted evaluation on datasets with high intrinsic dimension or non-uniform densities where LSH locality sensitivity is known to degrade. Because ZK-LSH-Shapley only proves the approximate protocol, any systematic bias in the underlying primitive directly undermines the 'on par' utility guarantee.

Authors: We agree that the current presentation relies on observed empirical gaps and would benefit from additional supporting analyses. In the revised manuscript we will report standard deviations and error bars on the AUROC figures by averaging over multiple independent runs of the LSH-Shapley procedure. We will also add a sensitivity analysis section that varies the number of hash functions, tables, and bucket widths, demonstrating stability of the approximation quality within the reported parameter ranges. Our 12 datasets already span a range of domains and dimensionalities (including image and high-dimensional feature data), and we will explicitly discuss results on subsets with higher intrinsic dimension or non-uniform distributions to address potential LSH degradation. While we acknowledge that tight concentration bounds on the collision-count estimator would provide stronger theoretical grounding, deriving such bounds for the composite Shapley-value estimator is non-trivial and outside the primary scope of this systems-oriented work; we will instead expand the discussion of empirical reliability and approximation limitations. These changes will better substantiate the practical 'on par' utility claim while preserving the focus on the end-to-end ZK system. revision: partial

Circularity Check

0 steps flagged

No circularity: ZK-Value is a new co-designed construction validated empirically against external baselines

full rationale

The paper's derivation chain consists of three independent components: (1) introduction of LSH-Shapley as an approximation primitive replacing pairwise distances with collision counts, (2) a tailored ZKP protocol encoding bucket histograms to reduce witness size, and (3) structural optimizations like super-oracle batching. These are presented as novel engineering choices, not derived from prior self-citations or self-definitions. Valuation quality claims rest on direct empirical comparison to exact KNN-Shapley and other baselines across 12 datasets (AUROC gap ≤0.033), not on any fitted parameter renamed as a prediction or on a uniqueness theorem imported from the authors' prior work. No equation reduces to its own input by construction, and the ZKP correctness is asserted via standard cryptographic assumptions rather than circular enforcement. The system is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The abstract does not provide details on specific free parameters, axioms, or invented entities used in the LSH-Shapley or ZKP components.

pith-pipeline@v0.9.0 · 5634 in / 1187 out tokens · 56255 ms · 2026-05-07T15:30:50.592247+00:00 · methodology

Review history (2 revisions) →

discussion (0)

Reference graph

Works this paper leans on

59 extracted references · 3 canonical work pages

[1]

Daniel Alabi, Sainyam Galhotra, Shagufta Mehnaz, Zeyu Song, and Eugene Wu
[2]

InCompanion of the 2025 International Conference on Management of Data

Privacy and security in distributed data markets. InCompanion of the 2025 International Conference on Management of Data. 775–787

2025
[3]

Nir Bitansky, Ran Canetti, Alessandro Chiesa, Shafi Goldwasser, Huijia Lin, Aviad Rubinstein, and Eran Tromer. 2017. The hunting of the SNARK.Journal of Cryptology30, 4 (2017), 989–1066

2017
[4]

Andrei Z Broder, Moses Charikar, Alan M Frieze, and Michael Mitzenmacher
[5]

InProceedings of the thirtieth annual ACM symposium on Theory of computing

Min-wise independent permutations. InProceedings of the thirtieth annual ACM symposium on Theory of computing. 327–336
[6]

Moses S Charikar. 2002. Similarity estimation techniques from rounding algo- rithms. InProceedings of the thiry-fourth annual ACM symposium on Theory of ZK-V alue: A Practical Zero-Knowledge System for Verifiable Data Valuation computing. 380–388

2002
[7]

Bing-Jyue Chen, Suppakit Waiwitlikhit, Ion Stoica, and Daniel Kang. 2024. Zkml: An optimizing system for ml inference in zero-knowledge proofs. InProceedings of the Nineteenth European Conference on Computer Systems. 560–574

2024
[8]

Lingjiao Chen, Paraschos Koutris, and Arun Kumar. 2019. Towards model-based pricing for machine learning in a data marketplace. InProceedings of the 2019 international conference on management of data. 1535–1552

2019
[9]

Mayur Datar, Nicole Immorlica, Piotr Indyk, and Vahab S Mirrokni. 2004. Locality- sensitive hashing scheme based on p-stable distributions. InProceedings of the twentieth annual symposium on Computational geometry. 253–262

2004
[10]

Shaleen Deep and Paraschos Koutris. 2017. QIRANA: A framework for scal- able query pricing. InProceedings of the 2017 ACM International Conference on Management of Data. 699–713

2017
[11]

Raul Castro Fernandez, Pranav Subramaniam, and Michael J Franklin. 2020. Data market platforms: Trading data assets to solve data problems.arXiv preprint arXiv:2002.01047(2020)

work page arXiv 2020
[12]

Amos Fiat and Adi Shamir. 1986. How to prove yourself: Practical solutions to identification and signature problems. InConference on the theory and application of cryptographic techniques. Springer, 186–194

1986
[13]

Ariel Gabizon and Zachary J Williamson. 2020. plookup: A simplified polynomial protocol for lookup tables.Cryptology ePrint Archive(2020)

2020
[14]

Sanjam Garg, Aarushi Goel, Somesh Jha, Saeed Mahloujifar, Mohammad Mah- moody, Guru-Vamsi Policharla, and Mingyuan Wang. 2023. Experimenting with zero-knowledge proofs of training. InProceedings of the 2023 ACM SIGSAC conference on computer and communications security. 1880–1894

2023
[15]

Amirata Ghorbani, Michael Kim, and James Zou. 2020. A distributional frame- work for data valuation. InInternational Conference on Machine Learning. PMLR, 3535–3544

2020
[16]

Amirata Ghorbani and James Zou. 2019. Data shapley: Equitable valuation of data for machine learning. InInternational conference on machine learning. PMLR, 2242–2251

2019
[17]

Aristides Gionis, Piotr Indyk, Rajeev Motwani, et al. 1999. Similarity search in high dimensions via hashing. InVldb, Vol. 99. 518–529

1999
[18]

S GOLDWASSER, S MICALI, and C RACKOFF. 1989. The knowledge complexity of interactive proof systems.SIAM journal on computing (Print)18, 1 (1989), 186–208

1989
[19]

Alexander Golovnev, Jonathan Lee, Srinath Setty, Justin Thaler, and Riad S Wahby
[20]

InAnnual International Cryptology Conference

Brakedown: Linear-time and field-agnostic SNARKs for R1CS. InAnnual International Cryptology Conference. Springer, 193–226
[21]

Binbin Gu, Juncheng Fang, and Faisal Nawab. 2025. PoneglyphDB: Efficient non-interactive zero-knowledge proofs for arbitrary sql-query verification.Pro- ceedings of the ACM on Management of Data3, 1 (2025), 1–27

2025
[22]

Meng Hao, Hanxiao Chen, Hongwei Li, Chenkai Weng, Yuan Zhang, Haomiao Yang, and Tianwei Zhang. 2024. Scalable zero-knowledge proofs for non-linear functions in machine learning. In33rd USENIX Security Symposium (USENIX Security 24). 3819–3836

2024
[23]

Ruoxi Jia, David Dao, Boxin Wang, Frances Ann Hubis, Nezihe Merve Gurel, Bo Li, Ce Zhang, Costas Spanos, and Dawn Song. 2019. Efficient task-specific data valuation for nearest neighbor algorithms.Proc. VLDB Endow.12, 11 (July 2019), 1610–1623. https://doi.org/10.14778/3342263.3342637

work page doi:10.14778/3342263.3342637 2019
[24]

Kevin Jiang, Weixin Liang, James Y Zou, and Yongchan Kwon. 2023. Opendataval: a unified benchmark for data valuation.Advances in Neural Information Processing Systems36 (2023), 28624–28647

2023
[25]

Pang Wei Koh and Percy Liang. 2017. Understanding black-box predictions via influence functions. InInternational conference on machine learning. PMLR, 1885–1894

2017
[26]

Paraschos Koutris, Prasang Upadhyaya, Magdalena Balazinska, Bill Howe, and Dan Suciu. 2015. Query-based data pricing.Journal of the ACM (JACM)62, 5 (2015), 1–44

2015
[27]

Yongchan Kwon and James Zou. 2022. Beta Shapley: a Unified and Noise-reduced Data Valuation Framework for Machine Learning. InInternational Conference on Artificial Intelligence and Statistics. PMLR, 8780–8802

2022
[28]

Succinct Labs. 2026. SP1. https://github.com/succinctlabs/sp1

2026
[29]

Alan Li, Qingkai Liang, and Mo Dong. 2024. Sparsity-aware protocol for zk- friendly ml models: Shedding lights on practical zkml.Cryptology ePrint Archive (2024)

2024
[30]

Xiling Li, Chenkai Weng, Yongxin Xu, Xiao Wang, and Jennie Rogers. 2023. Zksql: Verifiable and efficient query evaluation with zero-knowledge proofs. Proceedings of the VLDB Endowment16, 8 (2023)

2023
[31]

Zitao Li, Bolin Ding, Liuyi Yao, Yaliang Li, Xiaokui Xiao, and Jingren Zhou. 2024. Performance-based pricing for federated learning via auction.Proceedings of the VLDB Endowment17, 6 (2024), 1269–1282

2024
[32]

Jinfei Liu, Jian Lou, Junxu Liu, Li Xiong, Jian Pei, and Jimeng Sun. 2021. Dealer: An end-to-end model marketplace with differential privacy.Proceedings of the VLDB Endowment14, 6 (2021)

2021
[33]

Ruibang Liu, Minyu Chen, Dengji Ma, and Guoqiang Li. 2026. Bridging Privacy and Utility: A Verifiable Framework for Data Valuation via Zero-Knowledge Proofs.Cryptology ePrint Archive(2026)

2026
[34]

Tianyi Liu, Xiang Xie, and Yupeng Zhang. 2021. Zkcnn: Zero knowledge proofs for convolutional neural network predictions and accuracy. InProceedings of the 2021 ACM SIGSAC Conference on Computer and Communications Security. 2968–2985

2021
[35]

Carsten Lund, Lance Fortnow, Howard Karloff, and Noam Nisan. 1992. Algebraic methods for interactive proof systems.Journal of the ACM (JACM)39, 4 (1992), 859–868

1992
[36]

Qin Lv, William Josephson, Zhe Wang, Moses Charikar, and Kai Li. 2007. Multi- probe LSH: efficient indexing for high-dimensional similarity search. InProceed- ings of the 33rd international conference on Very large data bases. 950–961

2007
[37]

Konstantin D Pandl, Fabian Feiland, Scott Thiebes, and Ali Sunyaev. 2021. Trust- worthy machine learning for health care: scalable data valuation with the shapley value. InProceedings of the Conference on Health, Inference, and Learning. 47–57

2021
[38]

Jian Pei, Raul Castro Fernandez, and Xiaohui Yu. 2023. Data and AI model markets: Opportunities for data and model sharing, discovery, and integration. Proceedings of the VLDB Endowment16, 12 (2023), 3872–3873

2023
[39]

Li Peng, Jiayao Zhang, Yihang Wu, Weiran Liu, Jinfei Liu, Zheng Yan, Kui Ren, Lei Zhang, and Lin Qu. 2025. Reliable and Private Utility Signaling for Data Markets.Proceedings of the ACM on Management of Data3, 6 (2025), 1–27

2025
[40]

Wenjie Qu, Yijun Sun, Xuanming Liu, Tao Lu, Yanpei Guo, Kai Chen, and Jiaheng Zhang. 2025. {zkGPT}: An Efficient Non-interactive Zero-knowledge Proof Framework for {LLM} Inference. In34th USENIX Security Symposium (USENIX Security 25). 2045–2063

2025
[41]

Lloyd S Shapley et al. 1953. A value for n-person games. (1953)

1953
[42]

Ingredients

Rachael Hwee Ling Sim, Xinyi Xu, and Bryan Kian Hsiang Low. 2022. Data Valuation in Machine Learning:" Ingredients", Strategies, and Open Challenges.. InIJCAI. 5607–5614

2022
[43]

Haochen Sun, Tonghe Bai, Jason Li, and Hongyang Zhang. 2024. Zkdl: Effi- cient zero-knowledge proofs of deep learning training.IEEE Transactions on Information Forensics and Security20 (2024), 914–927

2024
[44]

Haochen Sun, Jason Li, and Hongyang Zhang. 2024. zkllm: Zero knowledge proofs for large language models. InProceedings of the 2024 on ACM SIGSAC Conference on Computer and Communications Security. 4405–4419

2024
[45]

Peng Sun, Liantao Wu, Zhibo Wang, Jinfei Liu, Juan Luo, and Wenqiang Jin
[46]

A profit-maximizing data marketplace with differentially private federated learning under price competition.Proceedings of the ACM on Management of Data2, 4 (2024), 1–27

2024
[47]

Mukund Sundararajan and Amir Najmi. 2020. The many Shapley values for model explanation. InInternational conference on machine learning. PMLR, 9269–9278

2020
[48]

Justin Thaler. 2022. Proofs, arguments, and zero-knowledge.Foundations and Trends®in Privacy and Security4, 2-4 (2022), 117–660

2022
[49]

Zhihua Tian, Jian Liu, Jingyu Li, Xinle Cao, Ruoxi Jia, Jun Kong, Mengdi Liu, and Kui Ren. 2022. Private data valuation and fair payment in data marketplaces. arXiv preprint arXiv:2210.08723(2022)

work page arXiv 2022
[50]

Riad S Wahby, Ioanna Tzialla, Abhi Shelat, Justin Thaler, and Michael Walfish
[51]

In2018 IEEE Symposium on Security and Privacy (SP)

Doubly-efficient zkSNARKs without trusted setup. In2018 IEEE Symposium on Security and Privacy (SP). IEEE, 926–943
[52]

Jiachen T Wang and Ruoxi Jia. 2023. Data banzhaf: A robust data valuation frame- work for machine learning. InInternational conference on artificial intelligence and statistics. PMLR, 6388–6421

2023
[53]

Jiachen Tianhao Wang, Yuqing Zhu, Yu-Xiang Wang, Ruoxi Jia, and Prateek Mittal. 2023. A privacy-friendly approach to data valuation.Advances in Neural Information Processing Systems36 (2023), 60429–60467

2023
[54]

Zhaoyu Wang, Pingchuan Ma, Zhantong Xue, Yanbo Dai, Zhenlan Ji, and Shuai Wang. 2025. Privacy-preserving and Verifiable Causal Prescriptive Analytics. Proceedings of the ACM on Management of Data3, 6 (2025), 1–27

2025
[55]

Chenkai Weng, Kang Yang, Xiang Xie, Jonathan Katz, and Xiao Wang. 2021. Mystique: Efficient conversions for {Zero-Knowledge} proofs with applications to machine learning. In30th USENIX Security Symposium (USENIX Security 21). 501–518

2021
[56]

Jinsung Yoon, Sercan Arik, and Tomas Pfister. 2020. Data valuation using re- inforcement learning. InInternational Conference on Machine Learning. PMLR, 10842–10851

2020
[57]

RISC Zero. 2026. RISC Zero. https://github.com/risc0/risc0

2026
[58]

Tianyu Zhang, Shen Dong, O Deniz Kose, Yanning Shen, and Yupeng Zhang. 2025. Fairzk: A scalable system to prove machine learning fairness in zero-knowledge. In2025 IEEE Symposium on Security and Privacy (SP). IEEE, 3460–3478

2025
[59]

RLC batch

Yiding Zhu, Hongwei Zhang, Jiayao Zhang, Jinfei Liu, and Kui Ren. 2024. Dat- aprice: An interactive system for pricing datasets in data marketplaces.Proceed- ings of the VLDB Endowment17, 12 (2024), 4433–4436. Zhaoyu Wang, Pingchuan Ma, Zhantong Xue, Yuguang Zhou, Qixin Zhang, Xiaoqin Zhang, and Shuai Wang A DATASETS AND PRE-PROCESSING Real-world panel.Th...

2024