Synthetic Data: Opening the data floodgates to enable faster, more directed development of machine learning methods

Many ground-breaking advancements in machine learning can be attributed to\nthe availability of a large volume of rich data. Unfortunately, many\nlarge-scale datasets are highly sensitive, such as healthcare data, and are not\nwidely available to the machine learning community. Generating synthetic data\nwith privacy guarantees provides one such solution, allowing meaningful\nresearch to be carried out "at scale" - by allowing the entirety of the machine\nlearning community to potentially accelerate progress within a given field. In\nthis article, we provide a high-level view of synthetic data: what it means,\nhow we might evaluate it and how we might use it.\n