Dynamic Fuzzy Machine Learning Model And Its Applications

Machine learning is a perpetual topic in the fields of Artificial Intelligence and Computer Science. At present, the more generally accepted definition of Machine Learning by most people is that of Simon:"If a system can through carry out some kind of process but to improve its performance, this was the study."The important points of this definition are: 1, Learning is a process; 2, Learning is to a system; 3, Learning can improve the performance of a system."Process","System"and"Improving performance"are three main factors of machine learning. We can find that machine learning system is a dynamic fuzzy system: for instance, the dynamic fuzzy character of learning process, the dynamic fuzzy character of system's variety and the dynamic fuzzy character of improving system's performance. Therefore, according to the dynamic fuzzy character in machine learning system, this paper proposed the Dynamic Fuzzy Machine Learning Model (DFMLM) and its related algorithms, the main contents of this paper are as follows:(1) Proposed the dynamic fuzzy machine learning model, then given the learning algorithm and the parameter learning algorithm and the maximum likelihood algorithm of Dynamic Fuzzy Machine Learning System (DFMLS).(2) Introduced the process control model of DFMLS and design of dynamic fuzzy learning controller, thus provided an effective method to solve the control problem of DFMLS.(3) The dynamic fuzzy relational learning algorithm was presented, further enriched the contents of Dynamic Fuzzy Machine Learning (DFML).(4) Given an example system: intelligent Gobang system.This paper studied the environment and learning algorithm of machine learning based on dynamic fuzzy sets, then proposed the algebra model and the geometry model of DFML, thus provided the theoretical foundation to solve the dynamic fuzzy problem of machine learning system. However, all the work is preliminary and much of them need advanced research. As future work, we plan to improve the learning algorithm and apply the dynamic fuzzy machine learning model to more fields, and so on.