MIIS

On December 17-18, 2023, the International Workshop on Mathematical Issues in Information Sciences (MIIS 2023) was held at The Chinese University of Hong Kong, Shenzhen. This two-day annual event aimed to bring together outstanding scientists, researchers, and engineers in the global field of data science to explore new ideas, theories, technologies, and applications in information science and big data. The goal was to promote the high-quality development of both the theoretical and industrial aspects of data science.

MIIS 2023 was hosted and organized by Shenzhen Research Institute of Big Data, co-sponsored by The Chinese University of Hong Kong, Shenzhen, Shenzhen International Center for Industrial and Applied Mathematics, Institute of Data and Information, Tsinghua Shenzhen International Graduate School, Xi'an Jiaotong University, National Key Laboratory of Radar Signal Processing, Shenzhen MSU-BIT University, National Health Data Institute, Shenzhen, Shenzhen Research Institute of Big Data Wuxi Innovation Center, and Guangdong Provincial Key Laboratory of Big Data Computing.

MIIS 2023 invited 19 world-renowned experts in the field of data science from academia, including Massimo Alioto, Zhiming Chen, Ambros Gleixner, Yongtao Guan, Haizhou Li, Xin Liu, Ya-Feng Liu, Xiaodong Luo, Yi Ma, Yurii Nesterov, Xipeng Qiu, Gianluca Setti, Jan Steinheimer-Froschauer, Defeng Sun, Takashi Tsuchiya, Gabriel Wittum, Jinchao Xu, Hai Yang and Raymond Yeung to share their academic insights. Over 200 scientists, researchers, engineers, and students attended the conference.

At the opening ceremony, Professor Zhi-Quan (Tom) Luo, Vice President of The Chinese University of Hong Kong, Shenzhen, and Director of Shenzhen Research Institute of Big Data, as well as a foreign member of the Chinese Academy of Engineering, delivered the welcome speech. He stated that the significant progress in information science and related technologies has had a crucial impact on various fields. With twelve years of history, MIIS has been committed to facilitating the sharing of the latest research findings among top global experts and scholars, promoting the exchange of cutting-edge research ideas, and addressing challenges in information and data science, big data analysis, and artificial intelligence.

Professor Xiaoping Zhang, Director of the Institute of Data and Information, Tsinghua Shenzhen International Graduate School, and a Fellow of the Canadian Academy of Engineering, then delivered a speech. He expressed that over the past decade, the MIIS conference has become a symbol in the field of data science, providing an important platform for the convergence and collision of diverse ideas for scientists, researchers, and engineers from both China and abroad.

Judi Jian, the Party Secretary of Shenzhen MSU-BIT University, emphasized that data science is becoming a key driving force for global technological innovation and social development. He highlighted that the MIIS conference would promote interdisciplinary communication and cooperation, injecting new vitality into the development of the field of data science.

Professor Xiaoping Wang, the Conference Chair of MIIS and Executive Director of Shenzhen International Center for Industrial and Applied Mathematics, stated that the MIIS conference encourages participants to delve into the forefront issues of data science in the era of artificial intelligence. He expressed enthusiastic anticipation for the conference's role in advancing technologies such as artificial intelligence and data science.

Professor Yurii Nesterov, recipient of the 2023 WLA Prize Laureates in Computer Science or Mathematics, shared a presentation titled "Optimization, the Philosophical Background of Artificial Intelligence." In his report, he provided a detailed overview of the application of optimization theory and methods in the development of artificial intelligence technology and within the field of computational mathematics.

The President of the Hong Kong Mathematical Society, Professor Defeng Sun from The Hong Kong Polytechnic University, presented a lecture titled "Nonsmooth Analysis and Sparse Optimization." In this presentation, he elucidated the significance of non-smooth analysis in large-scale sparse optimization. Professor Sun also introduced how to leverage non-smooth analysis to design efficient algorithms for solving large-scale machine-learning models.

Professor Yi Ma, Director of the Department of Computer Science at the University of Hong Kong, delivered a presentation titled "The Past, Present, and Future of Artificial Intelligence: from Black-Box to White-Box, from Open-Loop to Closed-Loop." In his talk, Professor Ma provided a more systematic and fundamental perspective on the practices of artificial intelligence over the past decade. He emphasized that optimization will offer a unified and clear explanation for the realization and application of artificial intelligence based on deep networks, including Convolutional Neural Networks (CNNs), Residual Networks (ResNets), and Transformers.

A renowned scholar in the international field of computational mathematics, Professor Jinchao Xu from KAUST, delivered a presentation titled "Deciphering the Curse of Dimensionality in Machine Learning." Through profound theoretical analysis, Professor Xu clarified and explained common misconceptions about high-dimensional problems in machine learning.

Professor Massimo Alioto from the National University of Singapore delivered a presentation titled "Green Technologies for Intelligent & Connected Systems at the Trillion Scale - Without Trillions of Batteries." Considering the adverse economic and environmental impacts of large-scale battery usage, Professor Alioto shared insights into a novel silicon system for wireless communication that operates from sensing and computation to communication without the need for batteries or any other energy storage, targeting intelligent systems with trillions of nodes.

Professor Yongtao Guan, Director of the Public and Judicial Big Data Laboratory at the Shenzhen Research Institute of Big Data, presented the topic "Group Network Hawkes Process." He introduced a model known as the Group Network Hawkes Process (GNHP), where the network structure is observed and fixed, to characterize the dynamic interaction processes of individuals within the network.

Professor Raymond Yeung from The Chinese University of Hong Kong presented a topic titled "Machine-Proving of Entropy Inequalities." In his presentation, Professor Yeung provided an overview of the development of machine proofs for entropy inequalities over the past 25 years. He introduced a geometric framework for entropy functions and discussed various conditions for the existence of entropy functions. Additionally, he explored how to construct corresponding entropy inequalities based on these conditions.

Professor Xin Liu from the Academy of Mathematics and System Science, Chinese Academy of Sciences presented the topic "Constraint Dissolving: A Powerful Tool for Riemannian Optimization." He introduced a constraint resolution method for a class of Riemannian manifold optimization problems. This method is capable of transforming Riemannian optimization problems into unconstrained optimization problems related to the corresponding constraint resolution functions.

Professor Ya-Feng Liu from the Academy of Mathematics and Systems Science, Chinese Academy of Sciences, presented the topic "One-Bit Precoding in Massive MIMO: Algorithm Design and Performance Analysis." He introduced the latest results concerning 1-bit precoding and addressed the precoding issues in both nonlinear and linearized precoding schemes. Additionally, a new negative l1 regularization method was introduced in the context of precoding problems.

Professor Zhiming Chen, an academician of the Chinese Academy of Sciences and a researcher at the Academy of Mathematics and System Science, Chinese Academy of Sciences, presented the topic "Arbitrarily High Order Finite Element Methods for Arbitrarily Shaped Domains with Automatic Mesh." He introduced a high-order non-matching finite element method for elliptic interface problems with hanging nodes on Cartesian grids. Moreover, he provided a natural high-order formulation design framework that does not rely on nonlinear element transformations.

Professor Gabriel Wittum from the College of Computer, Electronic, Mathematical Sciences, and Engineering at KAUST delivered a presentation titled "Parallel Adaptive Multigrid Methods for the Simulation of Processes from Science and Engineering." Based on the demand for large-scale numerical simulations, he introduced fundamental simulation strategies such as adaptivity, parallelism, and multigrid solvers. Additionally, he showcased the performance and efficiency of these strategies in various applications.

Professor Haizhou Li, Chief Scientist of the Shenzhen Research Institute of  Big Data and Academician of the Singapore Academy of Engineering, delivered a presentation titled "Large Language Model and Linguistics." He discussed the development of large language models from historical and theoretical perspectives, explored whether they could address linguistic issues, and showcased the latest advancements in GPT research conducted by Shenzhen Research Institute of Big Data and The Chinese University of Hong Kong, Shenzhen.

Professor Xipeng Qiu from the School of Computer Science at Fudan University presented the topic "Scientific Challenges of Large Language Models." He discussed issues related to large language models, including model architecture, reasoning ability, associative capability, and interpretability. Additionally, he put forth some forward-looking perspectives on future research directions.

Professor Jan Steinheimer-Froschauer from the Frankfurt Institute for Advanced Studies (FIAS) presented the topic "Big Data and AI in the Exploration of the Universe and Matter." He introduced new strategies employed by German scientists in fundamental research on the universe and matter, aiming to develop new information technologies and big data analysis methods. Additionally, he showcased some interdisciplinary research results from the collaborative efforts between FIAS and the Xidian-FIAS Joint Research Center.

Professor Hai Yang from the Hong Kong University of Science and Technology delivered a presentation titled "Mathematics, Economics and Artificial Intelligence for On-demand Mobility Services." He discussed the latest developments and current research issues in the on-demand ride-hailing market, covering topics such as competition, third-party platform integration, effective market regulation from a Pareto perspective, and the analysis of human mobility and network attributes using large-scale vehicle trajectory data.

Professor Gianluca Setti, Director of the College of Computer, Electronic, and Mathematical Sciences and Engineering (CEMSE) at KAUST, delivered a presentation titled "Techniques for TinyML: from Classical Pruning Methods to a New Neuron Paradigm for DNNs." He introduced a novel neuron structure based on the Multiplication-Max/Min (MAM) mapping-reduction paradigm. Professor Setti demonstrated that by leveraging this new paradigm, it is possible to construct naturally competitive sliced Deep Neural Network (DNN) layers with negligible performance loss.

Professor Takashi Tsuchiya from the Graduate School of Public Policy at the University of Tokyo (GRIPS) presented the topic "Perturbation Analysis and Algorithms for Singular SDPs with Nonzero Duality Gaps." Based on singular semidefinite relaxation problems that may have a nonzero duality gap, he utilized perturbation analysis to unveil the hidden continuity structure between the primal and dual problems.

Professor Ambros Gleixner from the Zuse Institute Berlin (ZIB) presented the topic "Verified Optimization of Mixed-Integer Programs without Numerical Errors." He introduced the latest advancements in high-performance mixed-integer programming solvers. This solver is not only unaffected by rounding errors but can also independently verify optimality conditions throughout the solving process. Additionally, he proposed a new analysis framework for verifying the accuracy of solvers constructed based on optimality systems.

Professor Xiaodong Luo, Director of the Center for General Software and Technologies of Big Data at  Shenzhen Research Institute of Big Data, presented the topic "Solver Technology and Its Applications." He showcased some of the work done in the field of solver development and highlighted the importance of integrating these technologies through examples in industries such as aviation, pharmaceuticals, and logistics.