Needle
A Generative AI-Powered Multi-modal Database for Answering Complex Natural Language Queries
Needle is an efficient and powerful text-to-image retrieval framework that significantly outperforms state-of-the-art baselines in multimodal information retrieval. This project represents a breakthrough in combining generative AI with database systems to solve complex natural language queries.
Key Achievements
- 200% improvement over OpenAI’s CLIP in mean average precision on complex natural language queries
- Advanced multimodal data management capabilities
- Integration of foundation models for enhanced retrieval performance
Technical Innovation
The system leverages cutting-edge generative AI techniques to bridge the gap between natural language queries and visual content, enabling more intuitive and accurate information retrieval across multimodal datasets.
Impact
This work has been submitted to ICLR 2025 and represents a significant advancement in the field of multimodal data management and information retrieval systems.
Research Context
This project is part of my Ph.D. research at the University of Illinois Chicago under the supervision of Dr. Abolfazl Asudeh, focusing on the intersection of generative AI and multimodal data management systems.