This project seeks to develop an array of methods for synthetically augmenting datasets that are of limited size and/or have missing entries. Furthermore, it seeks to adapt existing and/or develop new metrics for assessing the quality of the synthetically generated data and, thereby, provide quantifiable confidence in the models that are trained through them. While the topic of synthetic data generation is universal and generally applicable to any domain, in this project we will focus on datasets that relate to problem encountered in the area of hardware security. Specifically, we will apply the methods to three datasets consisting of actual measurements that the PI has access to through prior research efforts. These datasets correspond to three interesting and well-studied problems in trusted and assured microelectronics, namely detection of hardware Trojans, counterfeit ICs, and tampered wafers. In terms of synthetic data generation methods, we will investigate two different directions. The first direction will build upon a vast body of research in statistical distribution modeling through non-parametric Kernel Density Estimation (KDE), while the second direction will leverage contemporary results in generative AI, wherein Generative Adversarial Networks (GANs) are used to capture a distribution and distinguish between in-class and out-of-class elements. In addition to these two synthetic data augmentation methods and their variants, this project will also investigate metrics for assessing how well the synthetically generated data reflect the original distribution. Using classical statistical moments, such as mean, variance, skewness and kurtosis, as well as contemporary concepts, such as α-precision and β-recall, we will establish quantifiable confidence for the models trained with the synthetic data. Methods and metrics will be applied to and evaluated on the three existing hardware security datasets as well as any other dataset related to hardware assurance that the IAB members may be willing to provide.