Extending Fair Null-Space Projections for Continuous Attributes to Kernel Methods
Researchers have developed a new method to extend fairness projections for continuous attributes in machine learning, specifically for kernel methods. This approach, termed "continuous fairness," addresses a gap in existing literature which primarily focuses on discrete attributes. The novel technique involves a direct transformation of the kernel matrix, making it applicable to various models, including Support Vector Regression, and demonstrating competitive or improved performance on multiple datasets. AI
IMPACT Improves fairness metrics for continuous data, potentially enabling wider adoption of ML in sensitive applications.