news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

Shenzhen News Network, August 12, 2024 (Shenzhen Special Zone Daily reporter Li Li) On the afternoon of August 11, the theme forum "The application of large models in industrial manufacturing: Is it a real demand or blindly following the trend?" was held at the Shenzhen University Town Conference Center. Scholars and industry experts from the Greater Bay Area conducted in-depth discussions on the application value and future development direction of large model technology, as well as how to empower "industrial intelligent manufacturing".

According to reports, the forum was jointly launched by the Youth Computer Science Forum of the China Computer Society in Shenzhen and the Shenzhen Information Industry Association. It attracted representatives from 48 universities, enterprises and research institutions including Tsinghua University, Guangming Laboratory, CGN Research Institute, Skyworth, Lipu Intelligent Manufacturing, and Huaxing Optoelectronics. Qi Shuhan and Wu Yulin from Harbin Institute of Technology (Shenzhen) and Zhang Haiguang from Shenzhen Polytechnic University served as executive chairmen.

It is reported that the Greater Bay Area, especially Shenzhen, has a solid foundation for industrial manufacturing, including a large number of small and medium-sized enterprises, all of which are looking forward to leapfrogging development with the help of large-scale model technology. This forum aims to promote cooperation between academia and industry from both the supply and demand sides of industrial production, to provide advice and suggestions for large-scale model-enabled industrial upgrading, to achieve cost reduction and efficiency improvement for enterprises, and then to lead the Greater Bay Area and take the lead in promoting the popularization, application and innovative development of new quality productivity.

Liang Xiaojun, associate researcher of the Industrial Intelligence Team of Pengcheng Laboratory, said that the current industrial Internet does not adequately empower the underlying industrial control systems, resulting in limited advancement of industrial intelligence. The system as a whole faces development bottlenecks such as low computing power of industrial control systems, weak industrial data support capabilities, and bloated functions at the industrial manufacturing execution system layer. For this reason, industrial enterprises urgently need to break through the root technologies of independent intelligent control, form three key technology systems of intelligent control software and hardware, network applications, and testing and verification, and formulate a new industrial Internet standard system to lead the development of industrial Internet platforms.

Li Wenxian, vice president of the Sino-German Intelligent Manufacturing Institute of Shenzhen University of Technology, pointed out that large models have advantages such as improving efficiency, quality, and flexible production. Future research will focus on optimizing the computational efficiency of the model, improving the generalization ability of the model, enhancing data security protection measures, and formulating new ethical standards and regulations to deal with these emerging technologies. In addition, it will also explore the cross-application of robotics technology with other fields, such as using large models for more complex scenario simulation and prediction, so as to promote the practical application and social acceptance of robotics technology.

Faced with the need to implement big models in enterprises, some scholars have suggested that instead of relying on internal resources in isolation, it is better to use external resources and forces such as academia or research institutions to promote the implementation of big model technology in enterprises, thereby improving competitiveness. Some scholars also believe that there is no need to be too enthusiastic about big model technology, let alone deify it. Good data preparation may be the basis for the application of various technologies in the future.