Privately Fine-Tuned LLMs Preserve Temporal Dynamics in Tabular Data
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866908807865565184 |
|---|---|
| author | Rosenblatt, Lucas Liu, Peihan McKenna, Ryan Ponomareva, Natalia |
| author_facet | Rosenblatt, Lucas Liu, Peihan McKenna, Ryan Ponomareva, Natalia |
| contents | Research on differentially private synthetic tabular data has largely focused on independent and identically distributed rows where each record corresponds to a unique individual. This perspective neglects the temporal complexity in longitudinal datasets, such as electronic health records, where a user contributes an entire (sub) table of sequential events. While practitioners might attempt to model such data by flattening user histories into high-dimensional vectors for use with standard marginal-based mechanisms, we demonstrate that this strategy is insufficient. Flattening fails to preserve temporal coherence even when it maintains valid marginal distributions. We introduce PATH, a novel generative framework that treats the full table as the unit of synthesis and leverages the autoregressive capabilities of privately fine-tuned large language models. Extensive evaluations show that PATH effectively captures long-range dependencies that traditional methods miss. Empirically, our method reduces the distributional distance to real trajectories by over 60% and reduces state transition errors by nearly 50% compared to leading marginal mechanisms while achieving similar marginal fidelity. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_02766 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Privately Fine-Tuned LLMs Preserve Temporal Dynamics in Tabular Data Rosenblatt, Lucas Liu, Peihan McKenna, Ryan Ponomareva, Natalia Machine Learning Computation and Language Cryptography and Security I.2.6; I.2.7; D.4.6; K.4.1 Research on differentially private synthetic tabular data has largely focused on independent and identically distributed rows where each record corresponds to a unique individual. This perspective neglects the temporal complexity in longitudinal datasets, such as electronic health records, where a user contributes an entire (sub) table of sequential events. While practitioners might attempt to model such data by flattening user histories into high-dimensional vectors for use with standard marginal-based mechanisms, we demonstrate that this strategy is insufficient. Flattening fails to preserve temporal coherence even when it maintains valid marginal distributions. We introduce PATH, a novel generative framework that treats the full table as the unit of synthesis and leverages the autoregressive capabilities of privately fine-tuned large language models. Extensive evaluations show that PATH effectively captures long-range dependencies that traditional methods miss. Empirically, our method reduces the distributional distance to real trajectories by over 60% and reduces state transition errors by nearly 50% compared to leading marginal mechanisms while achieving similar marginal fidelity. |
| title | Privately Fine-Tuned LLMs Preserve Temporal Dynamics in Tabular Data |
| topic | Machine Learning Computation and Language Cryptography and Security I.2.6; I.2.7; D.4.6; K.4.1 |
| url | https://arxiv.org/abs/2602.02766 |