Digital Sandbox Initiative for Fintech Startups

In the processes of analysis and decision-making related to financial products, data-driven statistical and machine learning models are being used with increasing frequency. One of the main challenges in the development and training of such models remains the same - the issue of data quality and quantity. As part of the research into this issue, the National Bank of Georgia, through the involvement of its Financial and Supervisory Technology Development Department, is actively generating synthetic data using machine learning models and generative algorithms (Generative Adversarial Networks). The generated data has been successfully used for various tasks involving the hackathons.

Synthetic data is one of the most widely used methods in Privacy Enhancing Technologies, which allows for data sharing without the risk of disclosing real information or breaching confidentiality. This data only retains the statistical characteristics of real data, which is particularly important and valuable for statistical and machine learning models during their training and development stages.

In order to promote the development of financial technologies, support innovative ideas, and encourage competition in this field, the National Bank of Georgia is ready to cooperate with interested parties and, upon justified request, provide access to synthetic data. Within the framework of this initiative, both regulated and non-regulated entities will have the opportunity to use synthetic data-generated based on real data sources-for the purpose of testing their ideas, including the training and development of various statistical and machine learning models.