Chinanews.com, Shanghai News, April 14 (Shao Shibo, Xu Jing) Materials science + AI marks a new era of material innovation and discovery. Yu Dapeng, academician of the Chinese Academy of Sciences, president of the Shenzhen International Institute of Quantum Research, and chair professor of Southern University of Science and Technology, said in Shanghai that AI creation provides a strong tool support for solving the dilemma of material research and development in laboratory scientific research and the bottleneck of industrial application.
Yu Dapeng explained that on the one hand, AI opens the door to exploring a wider range of material possibilities compared to traditional materials research and development methods, significantly reducing the time and expense associated with material discovery, and on the other hand, material AI creation also faces challenges of credibility and effective implementation, and a series of issues such as ensuring data quality and identifying and mitigating potential biases in the data used to train AI systems need to be solved.
The School of Materials Science and Engineering of Shanghai Jiao Tong University and the Center for Future Materials Creation recently held the second phase of the "Youth Forum", inviting Yu Dapeng and young pioneer scientists in the field of "materials + AI" at Shanghai Jiao Tong University to brainstorm on the theme of "Where is the future of materials AI creation?".
Yu Dapeng told the young researchers and students present that although the mainland has made a lot of achievements in strengthening the creation of material AI in recent years, there is still a long way to go compared with world-class university research institutions and corporate giants. He encouraged the participating researchers to seize the opportunity of the era of revolutionary synthesis and exploration of materials, have the courage to carry out original, useful and extreme innovation in the field of "material innovation", and make unremitting efforts to realize the organic combination of man and machine of "AI (Artificial Intelligence) + HI (Human Intelligence)".
In addition to Academician Yu Dapeng, the forum also specially invited a number of young pioneer scientists from Shanghai Jiaotong University across colleges and interdisciplinary fields to exchange R&D experience and collide creation ideas around their self-developed material AI creation models. Sun Lizhen, Secretary of the Party Committee of the School of Materials Science and Engineering of Shanghai Jiao Tong University, delivered a welcome speech, Vice Dean Li Zhuguo and Distinguished Professor Dai Qing attended the forum, and Huang Fuqiang, Chair Professor of the School of Materials Science and Engineering of Shanghai Jiao Tong University and Director of the Center for Future Materials Creation, presided over the forum.
Wang Hong, chair professor of the School of Materials Science and Engineering of Shanghai Jiao Tong University and deputy director of the Center for Future Materials Creation, brought a report on "Line Station + AI: Future Materials Creation "Soft and Hard Expansion" Wings Fly Together", discussing the topic of "How to fully unleash the huge potential of AI?".
At present, the paradigm of materials science research is accelerating from trial and error to prediction, from hand-brain research to human-computer interaction. Wang Hong believes that data is the basic premise for the use of artificial intelligence, and AI empowers human knowledge creation to first solve the problem of "sufficient" and "easy to use" data.
At present, Shanghai Jiao Tong University is working with Shanghai Light Source, a major national scientific and technological infrastructure, to build the world's first special beam line for material genetic engineering and the first self-owned cable station of universities in China - "synchrotron radiation line station for material genome". As the main person in charge of the construction of the line station, Wang Hong introduced that the primary technical goal of the construction of the special line station for materials genome is to use general characterization technologies such as microbeam X-ray diffraction, fluorescence, small-angle scattering, wide-angle scattering, and absorption spectroscopy to build a material "data factory" featuring micro-area, scanning, fast, and high-throughput characterization and based on synchrotron radiation light sources. Wang Hong said that the "data factory" based on the hardware line station infrastructure aims to provide high-throughput data, and the "artificial intelligence" based on software machine reinforcement learning aims to mine multi-dimensional material structure-activity relationships, and the superposition of the two is expected to solve the "bottleneck" of data supply for material genetic engineering, and realize data sharing and recycling while accelerating the intelligent research and development of materials.
Li Jinjin, a researcher at the Artificial Intelligence and Microstructure Laboratory (AIMS-Lab) of the Department of Micro-Nano Electronics, School of Electronic Information and Electrical Engineering, Shanghai Jiaotong University, brought a report on "What is Alpha Mat.?". Majoring in physics, she organically integrates physics with materials science, life science, and information science, and leads the team to develop a new generation of material intelligent design model - Alpha Mat.
"Materials informatics is a multidisciplinary discipline that integrates materials science, computer science, artificial intelligence, physics, chemistry and other disciplines, however, such interdisciplinary talents are very scarce. Therefore, the development of a 'bridge' connecting materials science and artificial intelligence, closely connecting materials science and artificial intelligence and other disciplines, can not only promote materials science researchers to efficiently and effectively use AI tools to establish material prediction models, but also inspire more artificial intelligence scholars to innovate advanced algorithms and realize the application of technology. Li Jinjin said.
Li Jinjin leads the team to continuously optimize the Alpha Mat. Model. She introduced that the model has been released for 7 versions, and currently supports the whole process from "data acquisition→ data preprocessing→ feature engineering→ model establishment→ parameter optimization→ model evaluation→ result analysis", 26 AI models have been integrated, which can meet almost all modeling needs, 100 material data post-processing and analysis tools have been integrated, which can improve research efficiency, and more than 1 million material property databases have been >integrated and will be expanded in real time, no programming foundation is required, and it can be quickly applied in 5 minutes.
At the forum, Yao Zhenpeng, associate professor of the Center for Future Materials Creation of the School of Materials Science and Engineering of Shanghai Jiaotong University, who has participated in the construction of the industry's leading open quantum materials computing database OQMD, brought a report on "Embracing the Wave of Intelligence: An Overview of the Future Road of "Materials Creation". Zhu Hong, associate professor at the Joint Institute of Shanghai Jiao Tong University, presented a report entitled "AI + Materials Computing: How to Take the First Step in Material Learning". Zhu Hong introduced the status quo of the high-throughput computing platform independently developed by the team, and took the "research and development of stainless magnesium alloy materials and the design of new solid-state electrolyte materials" based on material genetic engineering methods as an example to illustrate the importance of reliable algorithms to achieve virtual experiments, active learning to reduce computing costs, and mechanisms to help material design first. (ENDS)
Editor: Xu Jing