As an important method of data analysis, machine learning technology has been widely used in academic and industrial circles. In particular, the development of cloud computing, logistics networks, big data, and quantum information has played an unprecedented role in promoting the globalization of machine learning. In modern data analysis, studying the technical level of computing may limit the whole process of scienti c research. As an analogy, analyzing the physical properties of a material or the physical properties of molecular structures is quite hard to achieve through calculations alone. The fundamental reason is the deep relationship between these data structures and the semantic uncertainty, which makes it di cult to complete certain tasks using only computing technology.
Dynamic fuzzy data (DFD) is one of the most di cult data types in the eld of big data, cloud computing, the Internet of Things, and quantum information. To investigate the deep structure of relations and the data semantic uncertainty of DFD, our research group has been studying this eld since 1994, and we have proposed various set, logic, and model system theories for uncertain datasets. To e ectively handle DFD, dynamic fuzzy sets, and dynamic fuzzy logic (DFL) are introduced into the machine learning eld, and the dynamic fuzzy machine learning theory framework is proposed.
Our work has been published in international journals and presented at inter- national conferences. The rst draft of “Dynamic Fuzzy Machine Learning” has been taught to master’s and doctoral students at Soochow University as an independent 54-hour course. Based on this, several revisions have been made. To meet the require- ments of the readers of this book, the course is published here over a total of seven chapters.
In the rst chapter, the dynamic fuzzy machine learning model is discussed. This chapter is divided into six sections. In the rst section, we de ne the problem. In the second section, we introduce the dynamic fuzzy machine learning model. In the third section, we study some related work. The fourth section presents the algorithm of the dynamic fuzzy machine learning system and the related process control model. In the fth section, we introduce the dynamic fuzzy relational learning algorithm. The sixth section summarizes the chapter.
The second chapter describes the dynamic fuzzy autonomous learning subspace learning algorithm. This chapter is divided into four sections. In the rst section, we analyse the current state of autonomous learning. In the second section, we present an autonomous learning subspace theoretical system based on DFL. In the third section, we introduce the autonomic subspace learning algorithm based on DFL. In the fourth section, we summarize this chapter.
The third chapter is devoted to fuzzy decision tree learning. This chapter is divided into six sections. In the rst section, we examine the current state of decision treelearning. In the second section, we study the dynamic fuzzy lattice decision tree method. In the third section, we discuss special attribute processing for dynamic fuzzy decision trees. In the fourth section, we study the pruning strategy for dynamic fuzzy decision trees. In the fth section, we describe some applications of dynamic fuzzy decision trees. The sixth section summarizes this chapter.
In the fourth chapter, we consider dynamic concepts based on dynamic fuzzy sets. This chapter is divided into seven sections. In the rst section, we analyse the relation between dynamic fuzzy sets (DFS) and concept learning. In the second section, we introduce the model of DF concept representation. In the third section, we model the DF concept learning space, and the fourth section describes the concept learning model based on the DF lattice. In the fth section, we present a concept learning model based on dynamic fuzzy decision tree. In the sixth section, we discuss some applications of the dynamic concept based on dynamic fuzzy sets and analyse their performance. In the seventh section, we summarize this chapter.
The fth chapter concentrates on semi-supervised multi-task learning based on dynamic fuzzy learning. This chapter is divided into six sections. The rst section introduces the notion of multi-task learning. In the second section, we describe the semi-supervised multi-task learning model. In the third section, we introduce a semi-supervised multi-task learning model based on DFS. In the fourth section, we introduce a dynamic fuzzy semi-supervised multi-task matching algorithm. The fth section extends this to a dynamic fuzzy semi-supervised multi-task adaptive learning algorithm. The sixth section summarizes this chapter.
In the sixth chapter, we study dynamic fuzzy hierarchy learning. This chapter is divided into seven sections. The rst section introduces the idea of hierarchical learning. In the second section, we describe the design of an inductive logic program. The third section discusses dynamic fuzzy hierarchical relational learning (HRL). In the fourth section, we study dynamic fuzzy tree HRL, and the fth section discusses dynamic fuzzy graph HRL. In the sixth section, we give some applications of dynamic concepts based on dynamic fuzzy sets and analyse their performance. The seventh section summarizes this chapter.
InChapter7,weconsideramulti-agentlearningmodelbasedonDFL.Thischapter is divided into ve sections. The rst section introduces multi-agent learning. In the second section, we introduce the agent mental model based on DFL. In the third section, we introduce the single agent learning algorithm based on DFL. The fourth section extends this idea to a multi-agent learning model based on DFL. In the fth section, we summarize this chapter. This book systematically introduces the relevant content of dynamic fuzzy learning. It can be used as a reference book for senior college students and graduate students as well as college teachers and scienti c and technical personnel involved in computer science, arti cial intelligence, machine learning, automation, mathematics, management science, cognitive science, nancial management, and data analysis. The text can also be used as the basis for a lecture course on dynamic fuzzy learning.
This book was designed and revised by Professors Li Fanzhang, Zhang Li, and Zhang Zhao. Huang Shuning, Cui Jingmei, Zoupeng, Luo Xiaohui, Wu Xinjian, Li Meixuan, Yin Hongwei, and Xu Xiaoxiang also assisted with the writing of the book. We are grateful for the wisdom and work of all teachers and students involved in the process. This book has also cited a large number of references, and it is our honour to express our deep gratitude to the authors of all references. Thanks to the National Nat- ural Science Foundation (61033013, 60775045, 61672364, 61672365), Soochow scholar program (14317360, 58320007) and the key subjects of Jiangsu province “Technology of computer science” and “Software engineering” for the support of this book.
Finally, I wish the book can bring happiness and inspiration to readers! Because this work is the rst attempt, combined with the author’s experience and knowledge being limited, if there is any improper place, please contact us. Contact method: E-mail: lfzh@suda.edu.cn,
