We are thrilled to announce that Professor Yuan Zhang has been awarded a National Science Foundation (NSF) research grant (amount 150K USD) for the collaborative project, "Statistical Foundations for Scalable and Robust Data Valuation." This foundational work tackles a critical challenge in modern artificial intelligence (AI): establishing transparent and reliable ways to measure the value of AI training data.

 

While modern AI systems rely heavily on massive datasets, the field currently lacks reliable mechanisms to properly credit or compensate the people and organizations contributing this data. Without accurate valuation, it is difficult to incentivize the sharing of the high-quality, socially beneficial data needed for future scientific breakthroughs. To solve this, Professor Zhang will lead research into developing new statistical and machine learning methods based on Shapley values, which were developed in the context of cooperative game theory to distribute rewards based on an individual’s unique contribution to a team's success.

 

Translating this game-theoretic concept into modern AI is highly complex. The research will focus on creating rules that factor in data originality and provenance, designing fast algorithms capable of handling massive datasets, and building new theories to measure data value reliably at a population level.