Correlation between Differential Privacy Metrics and LLM Reasoning Accuracy Degradation in Synthetic Tabular Data

Data serves as the fundamental foundation for advancing deep learning, particularly tabular data presented in a structured format, which is highly conducive to modeling. However, even in the era of LLM, obtaining tabular data from sensitive domains remains a challenge due to privacy or copyright concerns. Hence, exploring how to effectively use models like LLMs to generate realistic and privacy-preserving synthetic tabular data is urgent. In this paper, we take a step forward to explore LLMs for tabular data synthesis and privacy protection, by introducing a new framework HARMONIC for tabular

Research goal: How do differential privacy metrics in synthetic tabular data generation correlate with the degradation of downstream LLM reasoning accuracy on structured query tasks?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.7/10.