Elamur Research⌕
☰
home / papers / 2026 / From Rules to Predictions: Federated Tabular Learning with LLM Reasoning

From Rules to Predictions: Federated Tabular Learning with LLM Reasoning

Paper · 2026 · Afsaneh Mahanipour and Hana Khamfroush

Summary

The paper asks how to use LLM reasoning for tabular prediction when data sits in separate institutions, such as hospitals, and cannot be pooled or shown to a model. Serializing table rows into text for LLMs risks leaking records. It also handles anonymized features and exact numbers poorly, and it often fails to beat tree ensembles.

The authors propose a four-round federated protocol:

  1. Each client fits a local Random Forest and sends only the extracted decision rules (feature indices and thresholds).
  2. The server pools these rules, and an LLM is prompted once, with no fine-tuning, to produce a single refined global partitioning rule r*.
  3. Clients train leaf-specific residual correctors (a Gradient Net) on top of that rule.
  4. The correctors are merged with FedAvg.

The authors prove that this correction step strictly lowers expected loss for any positive step size. On medical and general tabular benchmarks, the method beats federated baselines and stays close to centralized models. This suggests that rules can serve as a privacy-preserving bridge between distributed data and LLMs.

Key points

  • The evaluation covers 12 datasets, seven of them clinical, including Heart Failure, Lung Cancer, Early Diabetes, Indian Liver, Pima Diabetes and Hepatitis C. Elsewhere the paper says eleven binary tasks, so the count is inconsistent.
  • The method is compared with FedAvg, Federated-FeatLLM and a Centralized-MLP upper bound. It is best on Accuracy, F1 and MCC for 6 of the 7 medical datasets.
  • Early Diabetes reaches 0.9747 accuracy. Hepatitis C reaches 0.9785 accuracy and 0.9086 MCC.
  • Against centralized LLM methods (LIFT, TP-BERTa, FeatLLM), the method tops Credit (0.762), Adult (0.859) and Jannis (0.6653). It trails TP-BERTa on Bank (0.898 vs 0.916) and ties it on Blood (0.761, slightly below FeatLLM's 0.768).
  • The average accuracy gap to the best centralized model is under 1%. The server only ever receives rule text and model parameters.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓University of KentuckyLexington · United States3 works · 260 citations

 Fields, topics and concepts

📚 Shelves 2The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 works

🧭 Topics 2Research subjects and languages the work is about.Language models & pretraining · 3,320 worksReasoning · 1,426 works

Cite

Cite as Afsaneh Mahanipour and Hana Khamfroush. 2026. From Rules to Predictions: Federated Tabular Learning with LLM Reasoning. BioNLP 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.bionlp-1.78

@inproceedings{mahanipour2026rules,
  title = {{From Rules to Predictions: Federated Tabular Learning with LLM Reasoning}},
  author = {Afsaneh Mahanipour and Hana Khamfroush},
  booktitle = {BioNLP 2026},
  year = {2026},
  doi = {10.18653/v1/2026.bionlp-1.78},
  url = {https://doi.org/10.18653/v1/2026.bionlp-1.78}
}

References

53 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. George J Annas. 2003. Hipaa regulations—a new era of medical-record privacy? find
  2. Samuel A Assefa, Danial Dervovic, Mahmoud Mahfouz, Robert E Tillman, Prashant Reddy, and Manuela Veloso. 2020. Generating synthetic data in finance: opportunities, challenges and pitfalls. In Proceedings of the first ACM international conference on AI in finance, pages 1–8. find
  3. Dominick L Frosch, David Grande, Derjung M Tarn, and Richard L Kravitz. 2010. A decade of controversy: balancing policy with evidence in the regulation of prescription drug advertising. American Journal of Public Health, 100(1):24–32. find
  4. Vadim Borisov, Tobias Leemann, Kathrin Seßler, Johannes Haug, Martin Pawelczyk, and Gjergji Kasneci. 2022. Deep neural networks and tabular data: A survey. IEEE transactions on neural networks and learning systems, 35(6):7499–7519. find
  5. Leo Breiman. 2001. Random forests. Machine learning, 45(1):5–32. find
  6. Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, and 1 others. 2020. Language models are few-shot learners. Advances in neural information processing systems, 33:1877–1901.
  7. Yury Gorishniy, Ivan Rubachev, Valentin Khrulkov, and Artem Babenko. 2021. Revisiting deep learning models for tabular data. Advances in neural information processing systems, 34:18932–18943. find
  8. Léo Grinsztajn, Edouard Oyallon, and Gaël Varoquaux. 2022. Why do tree-based models still outperform deep learning on typical tabular data? Advances in neural information processing systems, 35:507–520. find
  9. Sungwon Han, Jinsung Yoon, Sercan O Arik, and Tomas Pfister. 2024. Large language models can automatically engineer features for few-shot tabular learning. arXiv preprint arXiv:2404.09491. — [arXiv:2404.09491](https://arxiv.org/abs/2404.09491) find
  10. Tianqi Chen and Carlos Guestrin. 2016. Xgboost: A scalable tree boosting system. In Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining, pages 785– 794. find
  11. Kewei Cheng, Tao Fan, Yilun Jin, Yang Liu, Tianjian Chen, Dimitrios Papadopoulos, and Qiang Yang. 2021. Secureboost: A lossless federated learning framework. IEEE intelligent systems, 36(6):87–98. find
  12. Davide Chicco and Giuseppe Jurman. 2020. The advantages of the matthews correlation coefficient (mcc) over f1 score and accuracy in binary classification evaluation. BMC genomics, 21(1):6. find
  13. I Glenn Cohen and Michelle M Mello. 2018. Hipaa and protecting health information in the 21st century. Jama, 320(3):6–7. find
  14. Robert Detrano, Andras Janosi, Walter Steinbrunn, Matthias Pfisterer, Johann-Jakob Schmid, Sarbjit Sandhu, Kern H Guppy, Stella Lee, and Victor Froelicher. 1989. International application of a new probability algorithm for the diagnosis of coronary artery disease. The American journal of cardiology,… find
  15. Stefan Hegselmann, Alejandro Buendia, Hunter Lang, Monica Agrawal, Xiaoyi Jiang, and David Sontag. 2023. Tabllm: Few-shot classification of tabular data with large language models. In International conference on artificial intelligence and statistics, pages 5549–5581. PMLR. find
  16. Mikel Hernandez, Gorka Epelde, Ane Alberdi, Rodrigo Cilla, and Debbie Rankin. 2022. Synthetic data generation for tabular health records: A systematic review. Neurocomputing, 493:28–45. find
  17. Tin Kam Ho. 1998. The random subspace method for constructing decision forests. IEEE transactions on pattern analysis and machine intelligence, 20(8):832– 844. find
  18. Georg Hoffmann, Andreas Bietenbeck, Ralf Lichting-hagen, and Frank Klawonn. 2018. Using machine learning techniques to generate laboratory diagnostic pathways—a case study. Journal of Laboratory and Precision Medicine, 3(6). find
  19. Hans Hofmann. 1994. Statlog (german credit data). UCI Machine Learning Repository. https://doi.org/10.24432/C5NC77. — [doi:10.24432/c5nc77](https://doi.org/10.24432/c5nc77) ↗
  20. Tuan Dinh, Yuchen Zeng, Ruisu Zhang, Ziqian Lin, Michael Gira, Shashank Rajput, Jy-yong Sohn, Dimitris Papailiopoulos, and Kangwook Lee. 2022. Lift: Language-interfaced fine-tuning for non-language machine learning tasks. Advances in Neural Information Processing Systems, 35:11763–11784. find
  21. Vinu Ekanayake, Md Sultan Al Nahian, and Ramakanth Kavuluru. 2025. Mining social media for barriers to opioid recovery with llms. In Proceedings of the Second Workshop on Patient-Oriented Language Processing (CL4Health), pages 83–99. find
  22. Jerome H Friedman. 2001. Greedy function approximation: a gradient boosting machine. Annals of statistics, pages 1189–1232. find
  23. Noah Hollmann, Samuel Müller, Lennart Purucker, Arjun Krishnakumar, Max Körfer, Shi Bin Hoo, Robin Tibor Schirrmeister, and Frank Hutter. 2025. Accurate predictions on small data with a tabular foundation model. Nature, 637(8045):319–326. find
  24. MM Faniqul Islam, Rahatara Ferdousi, Sadikur Rahman, and Humayra Yasmin Bushra. 2019. Likelihood prediction of diabetes at early stage using data mining techniques. In Computer Vision and Machine Intelligence in Medical Image Analysis: International Symposium, ISCMM 2019, pages 113–125. Springer. find
  25. Sai Praneeth Karimireddy, Satyen Kale, Mehryar Mohri, Sashank Reddi, Sebastian Stich, and Ananda Theertha Suresh. 2020. Scaffold: Stochastic controlled averaging for federated learning. In International conference on machine learning, pages 5132–5143. PMLR. find
  26. Guolin Ke, Qi Meng, Thomas Finley, Taifeng Wang, Wei Chen, Weidong Ma, Qiwei Ye, and Tie-Yan Liu. 2017. Lightgbm: A highly efficient gradient boosting decision tree. Advances in neural information processing systems, 30. find
  27. Ron Kohavi. 1996. Scaling up the accuracy of Naive-Bayes classifiers: A decision-tree hybrid. In Proceedings of the Second International Conference on Knowledge Discovery and Data Mining (KDD), volume 96, pages 202–207. find
  28. Qinbin Li, Wu Zhaomin, Yanzheng Cai, Ching Man Yung, Tianyuan Fu, Bingsheng He, and 1 others. 2023. Fedtree: A federated learning system for trees. Proceedings of Machine Learning and Systems, 5:89– 103. find
  29. Tian Li, Anit Kumar Sahu, Ameet Talwalkar, and Virginia Smith. 2020a. Federated learning: Challenges, methods, and future directions. IEEE signal processing magazine, 37(3):50–60. find
  30. Tian Li, Anit Kumar Sahu, Manzil Zaheer, Maziar Sanjabi, Ameet Talwalkar, and Virginia Smith. 2020b. Federated optimization in heterogeneous networks. Proceedings of Machine learning and systems, 2:429– 450. find
  31. Wei-Yin Loh. 2011. Classification and regression trees. Wiley interdisciplinary reviews: data mining and knowledge discovery, 1(1):14–23. find
  32. Sérgio Moro, Paulo Cortez, and Paulo Rita. 2014. A data-driven approach to predict the success of bank telemarketing. Decision Support Systems, 62:22–31. find
  33. Jaehyun Nam, Kyuyoung Kim, Seunghyuk Oh, Jihoon Tack, Jaehyung Kim, and Jinwoo Shin. 2024. Optimized feature generation for tabular data via llms with decision tree reasoning. Advances in neural information processing systems, 37:92352–92380. find
  34. Christopher J O’Donnell and Roberto Elosua. 2008. Cardiovascular risk factors. insights from framingham heart study. Revista Española de Cardiología (English Edition), 61(3):299–310. find
  35. Liudmila Prokhorenkova, Gleb Gusev, Aleksandr Vorobev, Anna Veronika Dorogush, and Andrey Gulin. 2018. Catboost: unbiased boosting with categorical features. Advances in neural information processing systems, 31. find
  36. Bendi Venkata Ramana, M Surendra Prasad Babu, and NB Venkateswarlu. 2012. A critical comparative study of liver patients from usa and india: an exploratory analysis. International Journal of Computer Science Issues (IJCSI), 9(3):506. find
  37. Smita Raut, Shraddha Patil, and Gopichand Shelke. 2021. Lung cancer detection using machine learning approach. International Journal of Advance Scientific Research and Engineering Trends (IJASRET). find
  38. Afsaneh Mahanipour, Abdullah-Al-Zubaer Imran, and Hana Khamfroush. 2025. Federated reprogramming knowledge distillation for medical image classification. In International Conference on Medical Image Computing and Computer-Assisted Intervention, pages 143–152. Springer. find
  39. Nicola Rieke, Jonny Hancox, Wenqi Li, Fausto Milletari, Holger R Roth, Shadi Albarqouni, Spyridon Bakas, Mathieu N Galtier, Bennett A Landman, Klaus Maier-Hein, and 1 others. 2020. The future of digital health with federated learning. NPJ digital medicine, 3(1):119. find
  40. Afsaneh Mahanipour and Hana Khamfroush. 2025. Embedded federated feature selection with dynamic sparse training: balancing accuracy-cost tradeoffs. In 2025 International Joint Conference on Neural Networks (IJCNN), pages 1–8. IEEE. find
  41. Jack W Smith, James E Everhart, William C Dickson, William C Knowler, and Robert Scott Johannes. 1988. Using the adap learning algorithm to forecast the onset of diabetes mellitus. In Proceedings of the annual symposium on computer application in medical care, page 261. find
  42. Duncan McElfresh, Sujay Khandagale, Jonathan Valverde, Vishak Prasad C, Ganesh Ramakrishnan, Micah Goldblum, and Colin White. 2023. When do neural nets outperform boosted trees on tabular data? Advances in Neural Information Processing Systems, 36:76336–76369. find
  43. Yuan Sui, Mengyu Zhou, Mingjie Zhou, Shi Han, and Dongmei Zhang. 2024. Table meets llm: Can large language models understand structured table data? a benchmark and empirical study. In Proceedings of the 17th ACM International Conference on Web Search and Data Mining, pages 645–654. find
  44. Brendan McMahan, Eider Moore, Daniel Ramage, Seth Hampson, and Blaise Aguera y Arcas. 2017. Communication-efficient learning of deep networks from decentralized data. In Artificial intelligence and statistics, pages 1273–1282. Pmlr. find
  45. Paul Voigt and Axel Von dem Bussche. 2017. The eu general data protection regulation (gdpr). A practical guide, 1st ed., Cham: Springer International Publishing, 10(3152676):10–5555. find
  46. Luca Melis, Congzheng Song, Emiliano De Cristofaro, and Vitaly Shmatikov. 2019. Exploiting unintended feature leakage in collaborative learning. In 2019 IEEE symposium on security and privacy (SP), pages 691–706. IEEE. find
  47. Jianyu Wang, Qinghua Liu, Hao Liang, Gauri Joshi, and H Vincent Poor. 2020. Tackling the objective inconsistency problem in heterogeneous federated optimization. Advances in neural information processing systems, 33:7611–7623. find
  48. Jason Wei, Yi Tay, Rishi Bommasani, Colin Raffel, Barret Zoph, Sebastian Borgeaud, Dani Yogatama, Maarten Bosma, Denny Zhou, Donald Metzler, and 1 others. 2022. Emergent abilities of large language models. arXiv preprint arXiv:2206.07682.
  49. Qiang Yang, Yang Liu, Tianjian Chen, and Yongxin Tong. 2019. Federated machine learning: Concept and applications. ACM Transactions on Intelligent Systems and Technology (TIST), 10(2):1–19. find
  50. Hangting Ye, Jinmeng Li, He Zhao, Dandan Guo, and Yi Chang. 2025. Llm meeting decision trees on tabular data. arXiv preprint arXiv:2505.17918. — [arXiv:2505.17918](https://arxiv.org/abs/2505.17918) find
  51. I-Cheng Yeh. 2008. Blood transfusion service center. UCI Machine Learning Repository. https://doi.org/10.24432/C5GS39. — [doi:10.24432/c5gs39](https://doi.org/10.24432/c5gs39) ↗
  52. Yue Zhao, Meng Li, Liangzhen Lai, Naveen Suda, Damon Civin, and Vikas Chandra. 2018. Federated learning with non-iid data. arXiv preprint arXiv:1806.00582. — [arXiv:1806.00582](https://arxiv.org/abs/1806.00582) find
  53. Ligeng Zhu, Zhijian Liu, and Song Han. 2019. Deep leakage from gradients. Advances in neural information processing systems, 32. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/mahanipour-2026-rules-predictions-federated-tabular-learning/