ZKBoost: Zero-Knowledge Verifiable Training for XGBoost

2026-02-04

AI Summary

This work addresses the challenge of verifying the correctness of XGBoost training in privacy-sensitive settings without compromising data or model confidentiality. We propose zkPoT, the first zero-knowledge proof protocol tailored for XGBoost training, enabling a model owner to prove that training was correctly performed on a committed dataset while revealing no additional information. To achieve this, we design a fixed-point XGBoost implementation compatible with arithmetic circuits, a generic zkPoT proof template, and an efficient method for proving nonlinear fixed-point operations based on Vector Oblivious Linear Evaluation (VOLE). Experimental results on real-world datasets demonstrate that our approach incurs less than 1% accuracy degradation compared to standard XGBoost while enabling practical and efficient zero-knowledge verification of the training process.

Technology Category

• Machine Learning: Privacy
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories
• Data Mining & Knowledge Management: Representing, Reasoning, and Using Provenance, Trust

Application Category

• Security and Privacy: Data transparency and provenance
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Gradient boosted decision trees, particularly XGBoost, are among the most effective methods for tabular data. As deployment in sensitive settings increases, cryptographic guarantees of model integrity become essential. We present ZKBoost, the first zero-knowledge proof of training (zkPoT) protocol for XGBoost, enabling model owners to prove correct training on a committed dataset without revealing data or parameters. We make three key contributions: (1) a fixed-point XGBoost implementation compatible with arithmetic circuits, enabling instantiation of efficient zkPoT, (2) a generic template of zkPoT for XGBoost, which can be instantiated with any general-purpose ZKP backend, and (3) vector oblivious linear evaluation (VOLE)-based instantiation resolving challenges in proving nonlinear fixed-point operations. Our fixed-point implementation matches standard XGBoost accuracy within 1\% while enabling practical zkPoT on real-world datasets.

Problem

• zero-knowledge proof
• XGBoost
• verifiable training
• model integrity
• tabular data

Innovation

• zero-knowledge proof
• XGBoost
• fixed-point arithmetic
• zkPoT
• VOLE