Synthetic Text Generation with Differential Privacy: A Simple and Practical Recipe
This work addresses privacy concerns in data-driven products by offering a simple and practical method for generating synthetic text, though it is incremental as it builds on existing techniques.
The authors tackled the problem of generating high-quality synthetic text with strong privacy guarantees by fine-tuning a pretrained language model with differential privacy, resulting in synthetic text competitive in utility with non-private versions while providing strong protection against privacy leakages.
Privacy concerns have attracted increasing attention in data-driven products due to the tendency of machine learning models to memorize sensitive training data. Generating synthetic versions of such data with a formal privacy guarantee, such as differential privacy (DP), provides a promising path to mitigating these privacy concerns, but previous approaches in this direction have typically failed to produce synthetic data of high quality. In this work, we show that a simple and practical recipe in the text domain is effective: simply fine-tuning a pretrained generative language model with DP enables the model to generate useful synthetic text with strong privacy protection. Through extensive empirical analyses on both benchmark and private customer data, we demonstrate that our method produces synthetic text that is competitive in terms of utility with its non-private counterpart, meanwhile providing strong protection against potential privacy leakages.