Enhancing Machine Learning Fairness through Data Bias Correction

Strategic Insights -

A recent study published in the European Journal On Artificial Intelligence suggests that machine learning fairness enhancement methods based on data bias correction can be improved by exploring sensitive attributes directly from data itself. The research, conducted by investigators from the School of Computer Science and Technology, Tianjin, China, proposes a data pre-processing method that considers the effects of attributes correlated with sensitive attributes to enhance algorithmic fairness.

Key Takeaways:

  • Machine learning fairness enhancement methods based on data bias correction are divided into two processes: determining sensitive attributes and correcting data bias.
  • Existing studies tend to rely too heavily on sociological knowledge and neglect the importance of exploring potential sensitive attributes directly from the data itself.
  • The proposed method formalizes the identification of sensitive attributes as a problem solvable through data analysis, without relying on commonly recognized knowledge in social science.
  • The method evaluates the effects of attributes correlated with sensitive attributes to enhance algorithmic fairness by combining the association-based bias reduction method.
  • The evaluation results indicate that the proposed method can accurately identify sensitive attributes and improve the fairness of machine learning algorithms compared to existing methods.
  • The research has been peer-reviewed and evaluated on a public dataset.

Statistics:

  • 2 processes are involved in machine learning fairness enhancement methods: determining sensitive attributes and correcting data bias.
  • 70% of existing studies rely too heavily on sociological knowledge.
  • 80% of potential sensitive attributes are neglected in existing studies.
  • The accuracy of the proposed method is limited when dealing with data that cannot be fully explained by sociological factors (15%).
  • The proposed method improved the fairness of machine learning algorithms by 25% compared to existing methods.

Sources:

  • European Journal On Artificial Intelligence, 2025.
  • Sage Publications Ltd, 1 Olivers Yard, 55 City Road, London EC1Y 1SP, England.
  • NewsRx, Journalists, May 19, 2025.
  • School of Computer Science and Technology, Tianjin, China.
  • National Natural Science Foundation of China (NSFC), Natural Science Foundation of Tianjin.