After experiencing rapid advancements in the field of artificial intelligence and machine learning, synthetic data generation has emerged as a powerful solution to the persistent challenge of data scarcity. This trend is particularly critical in sensitive domains such as consumer credit lending, where fairness is paramount to prevent discriminatory outcomes and ensure equitable access to financial resources (Bartlett et al., 2022; Bhatore et al., 2020). While existing studies primarily assess synthetic data's capacity to improve algorithmic performance (Bansal et al., 2022), research on the fairness implications of synthetic data remains nascent. Consequently, understanding how synthetic data generation impacts fairness, suitable fairness evaluation methods, and strategies to mitigate bias during synthetic data generation is crucial for developing ethical and equitable machine learning systems.

At IWI, we strive to connect engaging academic research with practical real-world applications. Building upon insights from our recent paper: Not enough Data to be Fair? Evaluating Fairness Implications of Data Scarcity Solutions, your thesis will explore exciting and relevant questions through methods like systematic literature reviews, quantitative analyses, and practical experimentation with different synthetic data generation approaches.

• Empirical analysis of different synthetic data generation methods (e.g., CTGAN, Copula methods, diffusion models) and their specific impacts on fairness outcomes.
• Development and validation of novel fairness evaluation metrics suitable for assessing synthetic data quality and identifying biases in regression- as well as classification-based scenarios.
• Investigation of strategies and methodologies for mitigating bias during synthetic data generation, focusing on practical applications and measurable fairness improvements.
   

• The thesis can start immediately but should be completed within the next ±6 months.
• You possess a structured, conscientious, and result-driven work ethic.
• Your goal is to create a high-quality master thesis that contributes significantly to cutting-edge research.
   

• Close supervision through regular review meetings, feedback sessions, and collaborative discussions.
• Access to domain experts and datasets necessary for rigorous empirical evaluation.
• Opportunity to make meaningful contributions potentially leading to publication in respected academic journals. 

If you are interested, please reach out via email to mahei.li@unisg.ch. I look forward to discussing your thesis ideas and supporting you in achieving exceptional results.