Practical Anonymous Two-Party Gradient Boosting Decision Tree

2026-05-26

AI Summary

This work addresses the privacy-efficiency trade-off in secure GBDT training under vertical federated learning with mutually distrustful parties holding feature-partitioned data. Existing approaches rely on private set intersection (PSI) to align samples, inadvertently leaking shared record identifiers and compromising privacy. To overcome this limitation, we propose the first GBDT training framework that conceals record IDs by eliminating global alignment through an alternating dual-circuit PSI protocol and a state-sharing mechanism based on oblivious programmable pseudorandom functions. By integrating SIMD-enabled homomorphic encryption with RLWE-based optimizations, our method reduces ciphertext packing overhead by 50%. Experimental results demonstrate that the proposed scheme achieves training efficiency comparable to prior ID-leaking methods while fully preserving record ID privacy, and it is readily extensible to other vertical federated learning tasks.

Abstract

Structured data is well handled by gradient-boosted decision trees (GBDT), which are usually trained on vertically partitioned features across mutually distrustful parties. High speed and interpretability make GBDTs popular in finance and healthcare, where neural networks may fall short. Enabling secure computation for GBDTs poses unique challenges, requiring secure record alignment for comparison. Relying on private set intersection (PSI) is a de facto approach. Mistaking PSI for a safety measure actually exposes which record identifiers (IDs) are shared between the datasets. Although circuit-PSI could help, it is costly for generic uses. New ideas are needed to efficiently train in a "dark forest". Aiming to hide the IDs, we initiate the study of anonymous GBDT training on split data held by two parties. Dual circuit-PSI in our design lets the parties alternate as receiver to run pick-then-sum over local features. Via oblivious programmable pseudorandom functions, we propagate circuit-PSI outputs as shared state across runs. Avoiding universal alignment, we resolve the neglected dilemma that ID hiding incurs a cost that scales with domain size. Next, we halve the cost of ciphertext packing used to convert single-instruction multiple-data homomorphic encryption from (ring) learning with errors in prior secure GBDT (Usenix Security' 23) and related secure machine-learning computations. Comparative experiments show our protocol remains competitive with leaky approaches in efficiency. Enabling ID-hiding aggregation, our techniques can extend to other vertically partitioned analytics.

Problem

• anonymous GBDT
• vertical partitioning
• record identifier privacy
• secure computation
• private set intersection

Innovation

• anonymous GBDT
• dual circuit-PSI
• oblivious programmable PRF
• ciphertext packing optimization
• vertical federated learning