Collaborative LLM Numerical Reasoning with Local Data Protection
Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (41) · 34772-34780
DOI: 10.1609/aaai.v40i41.40779 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Min Zhang und 6 weiteren Autor:innen.
Abstract (Originalsprache)
Numerical reasoning over documents, which demands both contextual understanding and logical inference, is challenging for low-capacity local models deployed on computation-constrained devices. Although such complex reasoning queries could be routed to powerful remote models like GPT-4, exposing local data raises significant data leakage concerns. Existing mitigation methods generate problem descriptions or examples for remote assistance. However, the inherent complexity of numerical reasoning hinders the local model from generating logically equivalent queries and accurately inferring answers with remote guidance. In this paper, we present a model collaboration framework with two key innovations: (1) a context-aware synthesis strategy that shifts the query topics while preserving reasoning patterns; and (2) a tool-based answer reconstruction approach that reuses the remote-generated plug-and-play solution with code snippets. Experimental results demonstrate that our method achieves better reasoning accuracy than solely using local models while providing stronger data protection than fully relying on remote models. Furthermore, our method improves accuracy by 16.2% - 43.6% while reducing data leakage by 2.3% - 44.6% compared to existing data protection approaches.
Zitieren
APA
Zhang, M., Lu, Y., Zhou, Y., Xu, P., Cheong, L. L., Lu, C., & Wang, H. (2026). Collaborative LLM Numerical Reasoning with Local Data Protection. Proceedings of the AAAI Conference on Artificial Intelligence, 40(41), 34772-34780. https://doi.org/10.1609/aaai.v40i41.40779
Vancouver
Zhang M, Lu Y, Zhou Y, Xu P, Cheong LL, Lu C, et al. Collaborative LLM Numerical Reasoning with Local Data Protection. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(41):34772-34780. doi:10.1609/aaai.v40i41.40779
BibTeX
@article{Zhang2026,
author = {Zhang, Min and Lu, Yuzhe and Zhou, Yun and Xu, Panpan and Cheong, Lin Lee and Lu, Chang-Tien and Wang, Haozhu},
title = {Collaborative LLM Numerical Reasoning with Local Data Protection},
journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
year = {2026},
volume = {40},
number = {41},
pages = {34772--34780},
doi = {10.1609/aaai.v40i41.40779}
}
EUVIMED