Research

Basic information：

The intelligentdata analysis research team of the College of Mathematical Sciences is based on the construction of the doctoral program of mathematics and the key discipline of “Applied Mathematics” of Harbin Engineering University. The team consists of research directions including “granular computing and knowledge discovery”, “machine learning”, “decision and optimization”, and “maen-field game”. The team including 2 professors, 1 associate professor, 2 lecturers, 10 doctoral students and more than 20 master students. The teamhas undertaken a number of national, provincial and ministerial level and horizontal research projects.

Team leader:

DENG Tingquan,professor, doctoral supervisor of mathematics, control science and engineering. He was selected as the Youth Backbone of Academic Support Programs in Colleges and Universities in Heilongjiang Province, and was once an expert of the Information Department of the National Natural Science Foundation of China; Expert in the intergovernmental key special evaluation of the National Key Research and Development Plan of the Ministry of Science and Technology; Evaluation expert of “New Generation Artificial Intelligence” in Heilongjiang Province; Editor of Journal of Harbin Engineering University; Member of Particle Computing and Knowledge Discovery Committee of Chinese Artificial Intelligence Society; Vice chairman of Heilongjiang Industrial and Applied Mathematics Society.

Research interests: uncertainty modeling theory and method in data science, granular computing and knowledge discovery, machine learning and data mining, pattern recognition and artificial intelligence. He hosted three National Natural Science Foundation projects, a basic research program of the State Administration of Science, Technology and Industry for National Defense, and a number of national, provincial and ministerial level and horizontal research projects. He has published more than 110 academic research papers, and his research results have been cited more than 900 times by experts and scholars at home and abroad.

Team members:

WANG Changzhong, Professor, doctoral supervisor, distinguished professor of Northeastern University, distinguished professor of Liaoning Province. He is the director of Artificial Intelligence and Data Analysis Center of Bohai University, executive director of Liaoning Branch of Chinese Medical Mathematics Society, member of the special Committee of Basic Science of Artificial Intelligence Association, member of Multi-granularity and multi-scale Analysis Professional Committee of Chinese Society of Automation.

His research interests include machine learning, pattern recognition and data analysis methods based on particle computing. He hashostedmore than 10 projects such as the National Natural Science Foundation, Liaoning University Outstanding Talents Project, Liaoning Province Natural Science key Project, and published more than 60 papers.

LING Huanzhang, Ph.D.,associateprofessor, master supervisor,director of Basic academic Organization of Applied Mathematics Research Center, member of Educational Mathematics Committee of Chinese Society of Higher Education,member of China Society of Industrial and Applied Mathematics.

His research interests include the research of technology evaluation and decision, modeling and optimization under the background of JS. In recent years, he has cooperated with a research institute of Electric Science Group, provincial Academy of Sports and Social Sciences and other units andhosted8 scientific research projects,hosted4 central university special funds, participated in 3 National Natural Science Foundation funds, participated in 11 other projects, applied for 1 invention patent, published 1 book, and published 12 papers in academic journals and important academic conferences at home and abroad.

LI Qiang, PhD., master supervisor. Main research interests:application of variational and PDE methods in image and data analysis. It is also currently focused on learning algorithms.He published 3 SCI papers,and he was first author of 2 papers.

QIAOBin Fu, Lecturer,Ph.D.ofSophia University, Japan. At present, he has published 5 journal papers and 3 conference papers. He used to work as a research assistant at the Faculty of Science and Technology of Sophia University, Japan, participated in a research project commissioned by Toyota Motor East Fuji Research Institute.Heis currently in charge of a basic research fund for central universities.

His research interests include optimal control,mean-field game, and the application of multi-agent system cooperative control to practical problems

Main research directions of the team:

1. High-dimensional data mining feature extraction, target detection, recognition and tracking

(a) Semi-supervised feature extraction and clustering of high-dimensional data

For high-dimensional data, such as images, videos, etc., the key mathematical theoretical problems of unsupervised or supervised data reduction and clustering algorithms based on manifold learning, feature selection, feature extraction and other methods are studied, and effective and fast high-dimensional data mining algorithms are constructed to solve problems in practical engineering applications.

(b) Subspace learning for face recognition and image recovery

According to the distribution characteristics of high dimensional data such as video and image in low dimensional space, combining algebraic and geometry mathematical theory, as well as relevant methods in machine learning and digital image processing, more effective face recognition algorithm, moving object detection algorithm and image recovery algorithm are studied.

Face image recognition

Video target tracking

Subspace representation of missing content image recovery based on accelerated near-end gradient algorithm

2. Big data processing and knowledge discovery based ongranularcomputing

(a) Big data-oriented granular computing theory and method

Aiming at the hot issues such as topological structure, information graining, approximate approximation, uncertainty measurement and information fusion of multi-modal data under the background of big data, rough set and fuzzy set theories and methods are used to construct a granular computing model, explore data dimensionality reduction and pattern classification tasks, and study the theory and method of granular computing under large-scale data environment.

Multi-grain propagation

(b) Research on Three-way Decision making based on particle computing

Three-way Decision is a new method to simulate human thinking and solve complex problems in the field of cognitive computing. It is often combined with granular computing methodology. Fuzzy set, rough set, quotient space and other theoretical models are used to study complex problem solving, uncertainty problem expression and processing, massive data mining and fuzzy information processing.

Sequential three-waydecision method framework

(c) Outlier detection and fault diagnosis based on granular computing

Aiming at the problems of existing outlier mining methods and fault diagnosis, we explore and establishe the outlier detection method through nonlinear feature extraction and feature selection, unbalanced data clustering, classification and multi-level analysis from the perspective of granular computing.

Outlier mining results

3. Research on decision-making and optimization under the background of JS data

Decision-making and optimization problems in the context of JS data include path planning, task assignment, HS evaluation, target comprehensive capability evaluation, system evaluation, information support capability, target threat capability quantitative evaluation, fire distribution, complex network system construction and critical link analysis, bionic optimization and operational research technology, and specific prediction. The team uses modern multi-objective optimization and artificial intelligence algorithms to provide modeling analysis and optimization decisions for a series of problems, algorithms to solve many practical background problems, and provides relevant decision analysis.

Objective task assignment and intelligent obstacle avoidance

4. Research on the level set method of image segmentation

Image segmentation is a popular direction in recent years. Our research team explored and established a series of active contour image segmentation methods, including snake model, edge-based active contour and region-based active contour. Solve the segmentation problem of complex content images such as medical images and natural light images.

Human brain image segmentation results Rice grain image segmentation results

Color image segmentation results

5. Mean field game theory and its application

Mean Field Game (MFG) is a relatively new theoretical achievement in the field of system control and game theory. The main research object is the asymptotic property of Nash equilibrium point and the optimal control strategy of multiple agents in symmetric differential game system when the number of agents is approaching infinity. At present, some research results have been obtained: under the background of improving the utilization efficiency of road network traffic and optimizing energy consumption, the explicit solution of the quadratic mean field game problem for large-scale hybrid electric vehicles is given, and the distributed control law of this problem is obtained. A distributed charging control method based on mean field game collective consensus is proposed for large-scale electric vehicles.

Schematic diagram of control strategy structure

Speed and acceleration control effect Electric vehicle charging control effect

Academic Activities:

The intelligent data analysis research team takes the era of big data as the background, focuses on the current hot issues in the field of artificial intelligence, machine learning and other fields, explores the mathematical truth with a scientific and rigorous scientific research attitude, and trains outstanding talents for the national mathematical science and intelligent information processing construction.

Group discussionof differentresearchdirection

The teaminvitedinternationally renowned experts

The team jointly hosted the 2019 National Tianyuan Mathematics Summer training group photo

The team undertook the 2020 National Tianyuan Mathematics Summer Training (online)

Contact Us:

Interesteddoctors and visiting scholarsare welcome to join the research team!

Teachers and students interested in related research directions are welcome to join us!

The team enrolls1-2doctoral students and6-8graduateseveryyear.

Undergraduate students who want to join the team to study in advance, please contact

DENG Tingquan Professor:Deng.tq@hrbeu.edu.cn

FU Qiaobin Professor:fuqiaobin@hrbeu.edu.cn

If you don't want to be anAcnode among the vast number of scientific research workers,the intelligent data analysis research team is your clustering center!