Representative and equitable synthetic data: ML algorithms and working practices

Purpose and goal:

Our project’s goal is to develop a commercially available tool and a process for equitably representative synthetic data, based on industry needs. Our work will expand ML and synthetic data’s promise of representation to include the unique elements of the world’s edge cases that are often the most important.

Expected results and effects:

The effects of our project will include both technical solutions for representative synthetic data and market analysis to discover what alignment needs Swedish industry has for ML and synthetic data. The technical ML adjustments we will develop are going to be embedded in business process methods specifically tailored to Swedish industry’s synthetic data needs and working practices, ensuring bespoke data and alignment processes for increased quality of ML applications.

Approach and implementation:

Our project comprises three work packages: 1. Produce a method for determining the synthetic data bias needs of companies; 2. Create a technical solution to adjust the ML processes & produce less biased synthetic data; 3. Develop alignment strategies for Swedish industry to ensure equitable synthetic data practices.