Fair Model

Fair model research aims to create machine learning models that avoid perpetuating or amplifying biases against specific demographic groups, ensuring equitable outcomes across different populations. Current research focuses on developing and comparing various fairness-enhancing techniques, including algorithmic modifications (e.g., adaptive batch normalization, counterfactual reasoning), data preprocessing strategies (e.g., re-weighting, data augmentation), and post-processing methods (e.g., threshold adjustments), often within the context of federated learning. This field is crucial for mitigating the societal harms of biased AI systems in high-stakes applications like loan applications, healthcare, and criminal justice, promoting fairness and trust in AI technologies.