From December 30 to 31, 2023, in order to promote the integration and progress in the fields of statistics, operational research optimization and learning, the mathematical planning branch of the Chinese Operational Research Society, the China Field Statistics Research Association, the Beijing Institute of Applied Statistics, Beijing Jiaotong University, Beijing University of Technology, Capital Normal University, Renmin University of China The 8th International Seminar on Statistical Optimization and Learning, jointly sponsored by the Academy of Mathematics and Systems Science of the Chinese Academy of Sciences, was successfully held online and offline.

On the morning of December 30, the opening ceremony of the International Seminar on Online Statistical Optimization and Learning was held. Zhao Peng, Vice President of Beijing Jiaotong University, Guo Jianhua, President of China Field Statistics Research Association, President of Beijing Technology and Business University, Dai Yuhong, President of China Operation Research Society, Vice President of the Institute of Mathematics and Systems Science of the Chinese Academy of Sciences, attended the opening ceremony and delivered speeches. Harvard University, Pennsylvania State University Park, Carnegie Mellon University, National University of Singapore Experts and scholars from well-known institutions at home and abroad, such as Hong Kong Polytechnic University, the Chinese University of Hong Kong, Renmin University of China, University of the Chinese Academy of Sciences, Dalian University of Technology, Beijing Jiaotong University, attended the meeting. The opening ceremony was hosted by Yu Yongguang, Dean of School of Mathematics and Statistics, Beijing Jiaotong University.

Zhao Peng introduced the discipline construction of Beijing Jiaotong University, affirmed the important role of statistical optimization as an emerging discipline in data science, management science, transportation and other fields, and hoped that the seminar could promote the exchange of ideas, collaborative innovation, and promote the integration and progress of statistics and optimization, statistics and learning and other fields.

Guo Jianhua briefly introduced the development history of the Society, affirmed the great role played by the seminar series on statistical optimization and learning in promoting the integration of statistics, operational research optimization and learning, and stressed that in the data era, the intersection of multiple disciplines is a new trend and direction in the field of scientific research, hoping to promote the development of domestic statistical disciplines through the seminar.

Dai Yuhong believes that the integration of statistics and optimization disciplines also has a profound impact on operations research, and hopes to promote the development of applied mathematics and statistical optimization through disciplinary exchanges. He once again emphasized the purpose of the International Seminar on Statistical Optimization and Learning - to provide a platform for domestic and foreign scholars to exchange ideas and collide. At the same time, he hoped that the conference could help everyone find valuable research directions.

On the morning of December 30, the conference report was presided over by Wang Liqun, a professor of the University of Manitoba, and Xu Dachuan, a professor of Beijing University of Technology. Li Runze, a famous international statistician and Eberly Family Chair Professor of Pennsylvania State University, Yangtze River Chair Professor of the Ministry of Education, was invited to give a report. Li Runze introduced a high-dimensional matrix test method with M-estimation and the spectral unification theory of the high-dimensional Tyler's M-estimation when the potential distribution is in a more general situation. This method not only has good numerical performance and robustness in numerical experiments, but also has been applied in the empirical analysis of portfolio.

Kong Linglong, a professor at the University of Alberta in Canada, introduced the training robustness of distributed reinforcement learning in the face of noisy state observation. The report of Sun Ju, an assistant professor of the University of Minnesota, focused on the robustness of deep learning, proposed a software package for evaluating the robustness of deep learning models, and introduced a lightweight, universal selective classification method. The report of Associate Professor Ke Zheng of Harvard University focused on the text analysis and testing of high-dimensional polynomials, and gave the test statistics with asymptotic standard normal distribution under the null hypothesis and the optimal detection boundary of the test problem. Yuan Yancheng, an assistant professor of Hong Kong Polytechnic University, introduced some recent developments of convex clustering models, focusing on some methods that can enhance the clustering performance of convex clustering models by using dimension reduction technology of model structure and deep learning technology in appropriate ways.

In the afternoon of December 30, Cui Hengjian, a professor of Capital Normal University, Guo Tiande, a professor of University of Chinese Academy of Sciences, and Zhu Liping, a professor of Renmin University of China, presided over the report of the conference. Huang Jian, professor of data science and analysis in the Department of Applied Mathematics of Hong Kong Polytechnic University, was invited to give a report. Huang Jian explained the advantages of deep learning by considering the approximation ability of deep neural networks and the generalization error of deep learning. Kim Chuan Toh, a professor at the National University of Singapore, reported on the random Bregman gradient method and its application in deep learning. Li Xudong, associate professor of Fudan University, proposed an effective primal dual algorithm to solve minimax optimization problems with expected constraints. Tan Falong, associate professor of Hunan University, introduced a new method to test the parameter forms of mean and variance functions based on the weighted residual empirical process and its martingale transformation in the regression model. Dr. Wu Jiujing of Capital Normal University proposed a comprehensive modelless feature screening process based on Hellinger distance. Sun Defeng, professor of Hong Kong Polytechnic University, introduced some basic concepts of nonsmooth analysis and sparse optimization, and introduced how to use nonsmooth analysis to design efficient sparse nonsmooth Newton method and level set secant method.

On the morning of December 31, the report of the conference was presided over by Ding Chao, a researcher of the Institute of Mathematics and Systems Science of the Chinese Academy of Sciences, and Xing Wenxun, a professor of Tsinghua University. The conference invited Jin Jiashun, winner of IMS Tweedie Award and National Science Foundation Outstanding Youth Career Award, to report. Jin Jiashun introduced their latest research results, and used the collected journal citations and reference data to build the cited network model to analyze the research interests of scholars. Su Zhihua, an associate professor at the University of Florida, introduced a partial least square (PLS) based on envelope. Cui Ying, a professor at the University of California, Berkeley, introduced a new solver that can quickly solve large-scale optimization problems with super quantile constraints. Chen Kun, a professor at the University of Connecticut, introduced microbial research with sparse data characteristics, emphasizing the importance of feature selection. Nie Jiawang, a professor at the University of California, San Diego, introduced the geometric form of (weak) Pareto values, and how to use linear scalarization and Chebyshev table to quantitatively solve (weak) Pareto points and detect the existence of (weak) Pareto points.

At noon on December 31, the 2023 International Seminar on Statistical Optimization and Learning was successfully concluded. The closing ceremony was presided over by Professor Kong Lingchen of Beijing Jiaotong University. Professor Sun Defeng, President of the Hong Kong Mathematics Society and Head of the Department of Mathematics of the Hong Kong Polytechnic University, delivered a speech at the closing ceremony to congratulate the successful holding of the meeting. Sun Defeng reviewed the development of the International Seminar on Statistical Optimization and Learning and hoped that the seminar could promote the integration of various disciplines. Kong Lingchen expressed his gratitude to the speaker of the seminar and hoped that everyone would know more about it, establish a good development environment for statistical optimization and learning, and make contributions to the development of operational research and statistics in China.

The online and offline academic event was attended by scholars from the United States, Canada, Singapore and other countries and regions, as well as universities and research institutions across the country, with a total of 600 participants. At the beginning of the report of the conference, a four-day short-term course was held and researcher Zhang Xinyu from the Chinese Academy of Sciences was invited to give a lecture on "Model Average Frontier Selection", Professor Zhou Shenglong from Beijing Jiaotong University gave a lecture on "Sparse Optimization Second Order Algorithm", and Professor Lv Zhaosong from the University of Minnesota gave a lecture on "Machine Learning Optimization", Assistant Professor Zhang Yangjing of the Chinese Academy of Sciences teaches the course of "Optimization Algorithm for Large Scale Graph Models". More than 1500 people have attended short-term courses. Since 2013, the International Seminar on Statistical Optimization and Learning has been held for eight consecutive times. The series of conferences provided a high-level academic exchange platform for experts and scholars from different countries and regions, promoted the integrated development of statistics, operational research optimization, machine learning and other disciplines, and made positive efforts for the implementation of the national big data strategy.

Editor in charge: Li Xuemipiao

Reviewed by: Yuan Fang, Wang Ruixia