- Qifu Technology's (QFIN, Financial) research paper accepted at IJCAI 2025, with a competitive acceptance rate of 19.3%.
- The paper introduces TRIDENT, a framework using MLLM embeddings for enhanced fintech applications.
- TRIDENT shows potential in improving intelligent risk control and customer service within the fintech industry.
Qifu Technology (QFIN), in collaboration with Beijing Jiaotong University, has achieved a significant milestone with their research paper's acceptance at the prestigious International Joint Conference on Artificial Intelligence (IJCAI) 2025. With an acceptance rate of only 19.3%—selecting just 1,042 out of 5,404 submissions—this feat underscores the company's leading position in AI research.
At the core of their research is the TRIDENT framework, which utilizes Multimodal Large Language Model (MLLM) embeddings paired with attribute smoothing to advance Compositional Zero-Shot Learning. This innovative approach is designed to address inherent challenges in AI, such as background interference and limited semantic capture, thereby enhancing the performance across multiple datasets.
TRIDENT is set to revolutionize fintech operations by significantly improving two critical areas: intelligent risk control and customer service. By analyzing multimodal data, it promises more efficient detection of fraud patterns, while also enabling a sophisticated understanding of complex user inquiries to enhance customer support.
This development reflects Qifu Technology's strategic commitment to AI innovation and the fruitful outcomes of its investment in research and development, as well as its collaborations with academic institutions. As the company continues to explore and expand AI applications, these advancements are poised to contribute positively to the fintech sector and beyond.
