“Synthetic data generation techniques can generate training data using deep learning models such as the Generative Adversarial Networks (GANs). Such models can automatically discover and learn the statistical patterns and characteristics in an input dataset and then generate output datasets that share the same characteristics as the original dataset. Hence, synthetic data generation techniques can be used to rapidly produce a high volume of training data, bypassing the requirement to collect data from the real world. Furthermore, because it generates data from scratch, synthetic data techniques do not require labeling or feature engineering process, which is one of the most expensive, time-consuming, and error-prone activities in AI development. The labels are generated along with the dataset from the get go.”

- Wayne Eckerson in Business Monitoring Systems: Using Machine Learning to Analyze Business Metrics

June 15, 2020 (Report)