Synthetic-Powered Predictive Inference
This addresses the issue of data scarcity in predictive inference for machine learning practitioners, offering a novel method to enhance efficiency without distributional assumptions.
The paper tackles the problem of uninformative prediction sets in conformal prediction when calibration data are scarce by introducing Synthetic-powered predictive inference (SPI), which uses synthetic data and a score transporter to improve sample efficiency, resulting in substantially tighter prediction sets in experiments on image classification and tabular regression.
Conformal prediction is a framework for predictive inference with a distribution-free, finite-sample guarantee. However, it tends to provide uninformative prediction sets when calibration data are scarce. This paper introduces Synthetic-powered predictive inference (SPI), a novel framework that incorporates synthetic data -- e.g., from a generative model -- to improve sample efficiency. At the core of our method is a score transporter: an empirical quantile mapping that aligns nonconformity scores from trusted, real data with those from synthetic data. By carefully integrating the score transporter into the calibration process, SPI provably achieves finite-sample coverage guarantees without making any assumptions about the real and synthetic data distributions. When the score distributions are well aligned, SPI yields substantially tighter and more informative prediction sets than standard conformal prediction. Experiments on image classification -- augmenting data with synthetic diffusion-model generated images -- and on tabular regression demonstrate notable improvements in predictive efficiency in data-scarce settings.