Towards Fair In-Context Learning with Tabular Foundation Models

Keyphrases

• Sensitive Attribute 100%
• Foundation Models 100%
• Sample Selection 100%
• In-context Learning 100%
• Transformer-based 50%
• Minimal Impact 50%
• Structured Data 50%
• Prediction Accuracy 50%
• Learning Performance 50%
• Highly Sensitive 50%
• Benchmark Dataset 50%
• Enhancement Techniques 50%
• Equal Opportunities 50%
• Prediction Uncertainty 50%
• Protecting Groups 50%
• Gradient Boosted Trees 50%
• Method Correlation 50%
• Balanced Sample 50%
• Decorrelating 50%
• Attribute Group 50%
• Equal Representation 50%
• Demographic Parity 50%
• Attribute Prediction 50%
• Group Fairness Metrics 50%
• Equalized Odds 50%

Computer Science

• Sensitive Attribute 100%
• Foundation Model 100%
• Structured Data 50%
• Preprocessing 50%
• Reproducibility 50%
• Predictive Accuracy 50%
• Learning Performance 50%
• Gradient Boosting Tree 50%
• Equal Opportunity 50%
• Correlation Method 50%
• Attribute Group 50%