Data Valuation

Data valuation aims to quantify the worth of individual data points or entire datasets for machine learning, addressing the need for fair compensation and efficient data selection. Current research focuses on developing computationally efficient and robust methods, often leveraging Shapley values from game theory, influence functions, or optimal transport, to assess data contribution without requiring extensive model retraining. These advancements are crucial for building trustworthy AI systems, improving data marketplaces, and enabling more effective data management strategies across various applications.