Self Debiasing

Self-debiasing aims to mitigate biases—unfair prejudices learned from training data—in machine learning models, particularly large language models (LLMs) and other deep learning architectures. Current research focuses on developing methods to identify and reduce biases across various modalities (text, images, graphs), employing techniques like data augmentation, prompt engineering, and adversarial training within frameworks such as diffusion models and graph neural networks. Successfully addressing these biases is crucial for ensuring fairness, reliability, and ethical use of AI systems across diverse applications, ranging from healthcare and recruitment to social media and criminal justice.