CAAI cloud Forum (Jinan Station): Big Data and Intelligent Services, sponsored by CAAI, organized by CAAI Intelligent Service Committee and Jinan University, and co-organized by Shandong Association for Artificial Intelligence were successfully held on June 25, 2022. Well-known experts and entrepreneurs from industry, University and Research Institute were invited to exchange and discuss hot issues related to big data and intelligent services, so as to further help the technological innovation and application of the new generation of artificial intelligence. 519898 people watched this forum online.
This cloud forum was co-chaired by Wang Xiaoyang, Professor of School of computer science and technology of Fudan University, CAAI/CCF Fellow, deputy director of CAAI intelligent service special committee, and Du Junping, Professor of School of computer science of Beijing University of Posts and telecommunications, CAAI/CCF Fellow, director of CAAI intelligent service special committee. And National Science Fund for Distinguished Young Scholars, distinguished professor of Changjiang Scholars of the Ministry of education, director of CAAI intelligent creativity and digital art special committee, director of artificial intelligence collaborative innovation center of the Ministry of education, CAAI Fellow, Professor Zhuang Yueting of Zhejiang University, and National Science Fund for Distinguished Young Scholars, vice president of Jiangsu computer society, IETI Fellow, chief Professor, Professor Gengxin of executive vice president of Graduate School of Southeast University, and National Science Fund for Excellent Young Scholars, and leader of CCF university working group, Professor Tong Yongxin of Beijing University of Aeronautics and Astronautics, a foreign academician of the European Academy of Sciences, Professor Sun Maosong of executive vice president of the Institute of artificial intelligence of Tsinghua University, CAAI Fellow, made a special presentation. During the forum, the theme panel of "big data and intelligent service frontier" was also held.
The opening ceremony of the forum was presided over by Professor Wang Xiaoyang. He thanked the reporting guests who participated in the forum and introduced the sponsors, undertakers and co-organizers of the forum.
Professor Chen Yuehui, Vice President of Jinan University, delivered a speech, expressing a warm welcome to the experts and friends who participated in the forum. It is a great honor to be the organizer of this forum. The overall situation of Jinan University and the construction of the school's information discipline are introduced. The school actively builds a platform and focuses on condensing distinctive subject directions. I hope that all leaders and experts will continue to care and support Jinan University as always, and give more valuable suggestions for the reform and development of the school.
Professor Yin Yilong of Shandong University delivered a speech, introduced the development and membership of the Shandong Association for Artificial Intelligence, and expressed his gratitude to the experts for their strong support to the academic and industrial circles in Shandong Province and hope us to continue to support the Shandong Association for Artificial Intelligence and the development of big data and artificial intelligence-related fields in Shandong Province.
Professor Du Junping, director of the CAAI Intelligent Service Committee and Beijing University of Posts and Telecommunications, delivered a speech and introduced the basic situation of the Intelligent Service Committee. As the largest branch of the CAAI, the special committee vigorously promotes the innovation of intelligent service theory, technology and application. This forum is the continuation and expansion of the "CAAI Cloud Forum (Beijing Station): Big Data and Artificial Intelligence" in 2020. It is hoped that through the holding of this forum, big data and artificial intelligence technologies can be further promoted to empower intelligent services and help the Innovation and application of new generation of artificial intelligence technology.
Prof. Wang Xiaoyang hosted a panel discussion on the theme of "Frontiers of Big Data and Intelligent Services". Panel guests included Prof. Sun Maosong, Prof. Zhuang Yueting, Prof. Geng Xin, Prof. Yin Yilong, Prof. Tong Yongxin and Prof. Du Junping. And the exchanges and discussions were held on the theme of "Frontiers of Big Data and Intelligent Services".
Professor Sun Maosong of Tsinghua University believes that the combination of big data and intelligent services is a very good direction. He reviewed the development of the artificial intelligence wave and analyzed the role of natural language processing technology in the field of artificial intelligence. At present, technologies such as speech recognition, speech synthesis, machine translation, search engines, and language models have provided intelligent services for daily life. The latest language large model provides a good opportunity for the next generation of intelligent services. In addition to the existing intelligent service scenarios, we should think about what new intelligent services can be done, and how to give full play to the value and role of the large model.
Professor Zhuang Yueting from Zhejiang University conducted an analysis on multiple knowledge expressions, two-wheel drive of knowledge and data, and large-scale pre-training models. How to combine knowledge with deep neural network for multiple knowledge representation is very important. The analysis is carried out around the two-wheel drive of knowledge and data, and the knowledge-driven interpretability is analyzed on the basis of data-driven. Large-scale pre-training models are the basis for intelligent services of many technology enterprises. How to better develop intelligent services based on big data and using large-scale pre-training models is a key issue.
Professor Geng Xin from Southeast University analyzed the role of natural language processing, computer vision, big data and other technologies. Pre-trained large models based on big data solve many real-world problems. The marginal effects of the model are explored around the increase in data volume and network size. Training a high-quality model requires a large amount of data and expensive GPU resources for support. How big data and big models can meet the needs of high efficiency and low cost is an important problem that needs to be solved.
Professor Yin Yilong of Shandong University discussed the cultivation of big data and artificial intelligence talents in my country, which is divided into three stages. The first stage is the catch-up period, and a long-term effort and catch-up have been carried out in terms of the construction of the teaching staff and the reform of the training mechanism. The second stage is the "parallel run" period, and there is still a big or small gap between the overall training level and the international famous schools. The third stage is the lead period. This is the future, but it is not far away. How to cultivate strong leaders who better meet the requirements of the "Four Aspects" and have the ability to "freely switch" between academia and industry in the future The outstanding talents developed by the international academia and industry are important issues worthy of our in-depth consideration.
Professor Tong Yongxin from Beihang University discussed the development of intelligent service and big data technology and theory from the perspective of application, and discussed the concept of swarm intelligence and key technical issues. The focus of Internet-based swarm intelligence is to use the Internet to allocate resources reasonably, to attract, aggregate and manage large-scale Internet participants, and to complete a task collaboratively, thereby surpassing individual intelligence. A discussion was launched around federated learning, a popular research topic in the field of swarm intelligence, which is an extension of swarm intelligence. How to protect the privacy of data, relieve the pressure of communication overhead in the communication process, and do a good job of safe and efficient intelligent services are issues worthy of attention.
Professor Du Junping from the School of Computer Science of Beijing University of Posts and Telecommunications introduced that the research group has realized the intelligent analysis and processing of scientific and technological resources across fields, disciplines, media, and time and space with the support of the National Key R&D Program; Introduced the major project of the National Natural Science Foundation of China launched this year, "Basic Research on Efficient Learning of Complex Big Data and Public Safety Application", established an efficient representation model of complex big data, carried out high-level reasoning on the semantics expressed by big data, and realized public safety emergencies. Event discovery and evolution analysis, study the federated learning method that balances and optimizes data quality and learning efficiency, and realizes multi-party collaborative public security emergency characterization and situation prediction.
In the report section of the forum, four experts made wonderful reports respectively, chaired by Professor Wang Xiaoyang and Professor Du Junping respectively. The theme of the report by Professor Zhuang Yueting of Zhejiang University is "Intelligent Service: Let Big Data Shine" and "Hot". The term "big data" has been around for several years. The importance of big data is equated with future energy, electricity, etc. China's "New Generation Artificial Intelligence Development Plan" also puts forward a new understanding of "big data intelligence". Re-understanding and thinking about basic issues such as big data, knowledge expression and knowledge acquisition, and discussing current research trends, such as multiple knowledge expression, dual-wheel drive of knowledge and data, (super) large-scale pre-training models, cross-media intelligence Combining with others, it introduces the current generalizable knowledge computing engine and other intelligent services and technologies, and introduces the application of big data intelligence in the digital economy fields such as digital city governance, industrial chain early warning and precise investment promotion.
The theme of the report by Professor Geng Xin of Southeast University is "The 'Genes' of Machine Learning: Let Machines Learn Like Humans". One of the fundamental drivers of this AI boom is the development of modern machine learning, especially deep learning technology. Deep learning requires a lot of training data and computing resources. However, humans can often learn a new concept quickly with a small number of samples. This is because the newborn brain is not random, and the evolution of human beings over thousands of years has initialized the newborn brain through genes. Inspired by this, he proposed the "gene" of machine learning - Learngene. A new learning framework based on Learngene promises to be a game changer in the current deep learning field: now only a few "model suppliers" need to perform large-scale training on open-world tasks, while a task-specific "client" only needs to be trained from " Buy a Learngene from "Model Supplier" and initialize your own lightweight model, you can quickly adapt to your target task with a small number of samples, just like human learning.
The theme of Professor Sun Maosong's report from Tsinghua University is "Language Large Model and Its Typical Intelligent Service Application". The language large model is the dual commanding heights of the academic frontier and the technological frontier in the international artificial intelligence field in recent years. It has greatly improved the performance of many important tasks in the intelligent processing of text information, and its influence has rapidly expanded from the text field to multi-modality. State-of-the-art computing and even the multidisciplinary field of artificial intelligence for science. Firstly, it expounds the characteristics of the language large model and its important position as a "soft" infrastructure for intelligent information processing. Through several typical cases, it evaluates and discusses the typical new applications of intelligent services that can be empowered under its support, including intelligent writing ( Application genre), intelligent writing (creation), grammar correction, reverse dictionary and real-virtual world interaction with intelligent question answering as the core and multi-modal technology integration.
Professor Wang Xiaoyang made a summary of this cloud forum, looking forward to the development of more intelligent service-related technologies to contribute to the development of science and technology and social progress. CAAI Cloud Forum (Jinan Station): Big data and intelligent services have come to a successful conclusion.