AI-Powered Food Safety Detection
Using LLMs to Identify Food Hazards from Text
This research explores advanced text-based detection of food hazards with a focus on handling rare hazard categories through machine learning.
- Successfully leveraged synthetic data generated by large language models to improve rare hazard detection
- Developed a two-tier classification system: identifying hazard categories and providing fine-grained product-specific labels
- Demonstrated effective methods for addressing long-tail distribution challenges in food safety monitoring
This work has significant medical implications by enhancing public health protection through automated screening of web content for potential food hazards, enabling faster identification of emerging health risks.