Big data is changing the way we live and think. Basic mathematics highlights tremendous power. Optimized modeling is one of the powerful tools for decrypting data problems. On the afternoon of June 10th, Academician Yuan Yaxiang of the Chinese Academy of Sciences gave a scientific report entitled “Big Data and Optimization Methods” in the first lecture hall of the conference center. The report was hosted by Vice President Chen Hong.
Academician Yuan Yaxiang first analyzed the background of the actual problems of big data and the related mathematics fields involved, and pointed out that the amount of data in the social and economic life of the big data era has grown exponentially. No matter from the aspects of data collection, storage and analysis, it is necessary to establish a suitable mathematical model to analyze and process the data. Academician Yuan also introduced the current internationally-recognized optimization problems such as compression sensing, matrix integrity, classification problems, gradient methods, stochastic gradient methods, alternating direction methods, and augmented Lagrangian methods. He further gave some important inspirations for the optimization method. First, the classic and simple method has a good development prospect in the context of today's big data. Second, in the context of big data, the problems faced generally have a special structure, and the method of divide and conquer can be used to continuously simplify the problem and achieve the goal of optimization. Third, the basic mathematics is huge, and the development of data science cannot be separated from basic mathematics.
Academician Yuan finally interacted with the teachers and students on the topic of local optimum and global optimality. The wonderful report of Academician Yuan was enthusiastically resonated and resonated by the teachers and students.
