Yiming Chen
Algorithm Team Leader, Hundsun Technologies Research Institute
Yiming Chen, Postdoctoral Fellow of Institute of Automation, Chinese Academy of Sciences, Beijing & Visiting Scholar of McGill University, Canada, his main research interests are NLP, computer vision, edge computing and so on. He won the champion of CCF (China Computer Federation) Action Recognition Algorithm Competition 2022, the champion of Kaggle International Algorithm Image Caption Matching Competition, and was invited to make a special sharing at ICLR2022, the top conference of Artificial Intelligence Class A. He has been invited to participate in one national level oceanic project, and has completed one national level oceanic project. He has participated in and completed one national level ocean project. He has published more than 10 international top academic papers and more than 20 authorized patents. He has served as a tutor for special projects at Tsinghua University, and as an extracurricular tutor for Northeastern University, New Oriental, Taitze Technology, and many other colleges and universities (companies).
Topic
Technical bottlenecks and breakthroughs in the application of large models in the financial field
This talk focuses on the development and application challenges of big models. In recent years, big models have made significant progress in natural language processing, computer vision and other fields by virtue of their powerful learning and processing capabilities, showing great application potential. However, in the process of practical application, big models face many problems. For example, in ultra-long form Q&A, due to the complex structure of form data and large amount of information, it is difficult for the big model to accurately understand and answer related questions. In non-training situations, how to stimulate the potential ability of the big model so that it can flexibly cope with various new tasks is still a difficult problem to be solved. In addition, the high cost in application development limits the wide application of big models; therefore, low-cost application development techniques, such as optimizing resource allocation and rationally selecting algorithms, are important for promoting the popularity and application of big models. Exploring ways to solve these problems will help enhance the practicality and economic benefits of big models and promote their application and development in more fields.
