Machine learning is a multi-domain interdisciplinary subject involving statistics, probability theory, algorithm complexity theory, approximation theory, convex analysis, and other disciplines. It specializes in how computers simulate or realize human learning behaviors to acquire new skills or knowledge, and reorganize existing knowledge structures to continuously improve their performance.
Definition
Machine learning is a multidisciplinary interdisciplinary major, covering knowledge of complex algorithms, statistics, approximate theory, and probability theory. It uses computers as tools and is carried out to simulate real-time human learning techniques, and transforms existing content into knowledge structures. Division to effectively increase learning efficiency.
Machine learning has the following definitions:
- Machine learning is the use of data or past experience to optimize the performance criteria of computer programs.
- ML(Machine learning) is a science of artificial intelligence. The main research object in this field is artificial intelligence, especially how to better the performance of specific algorithms in observed learning.
- Machine learning is the study of computer algorithms that can be automatically better through experience.
Development Path
Machine learning has actually been around for decades or, arguably, centuries. Dating back to the 17th century, Bayesian and Laplace’s derivation of least squares and Markov chains formed the tools and foundations of widely used machine learning. From 1950 (Alan Turing proposed to build a learning machine) to early 2000 (with practical applications of deep learning and more recent advances, such as AlexNet in 2012), machine learning has made great progress.
Since the study of machine learning in the 1950s, the research approaches and goals in different periods are different and can be divided into four stages.
First Stage
The first stage was from the mid-1950s to the mid-1960s, and this period mainly studied “learning with or without knowledge”. This kind of method mainly studies the execution ability of the system. During this period, the data fed back by the system is mainly detected by changing the environment of the machine and its corresponding performance parameters. It is like giving the system a program. By changing their free space function, the system will be affected by the program and change its own Organization, and finally, this system will choose an optimal environment to survive. The most representative study in this period is Samuel’s chess program. But this machine-learning approach is far from meeting human needs.
Second Stage
The second stage is from the mid-1960s to the mid-1970s. This period mainly studies the implantation of knowledge in various fields into the system. The purpose of this stage is to simulate the process of human learning through machines. At the same time, the knowledge of graph structure and its logical structure is used for systematic description. In this research stage, various symbols are mainly used to represent machine language.
When conducting experiments, researchers realize that learning is a long-term process. In this system environment, it is impossible to learn more in-depth knowledge, so researchers add the knowledge of various experts and scholars into the system, and practice has proved that this method has achieved certain results. The representative work in this stage is Hayes-Roth and Winson’s systematic approach to structure learning.
Third Stage
The third stage, from the mid-1970s to the mid-1980s, is called the revival period. During this period, people have expanded from learning a single concept to learning multiple concepts, exploring different learning strategies and learning methods, and have begun to combine learning systems with various applications at this stage with great success. At the same time, the demand for knowledge acquisition by expert systems has greatly stimulated the research and development of machine learning.
After the emergence of the first expert learning system, the example inductive learning system has become the mainstream of research, and automatic knowledge acquisition has become the research goal of ML applications. In 1980, the first International Symposium on Machine Learning was held at Carnegie Mellon (CMU) in the United States, marking the rise of ML research around the world. Since then, ML has been widely used. In 1984, more than 20 artificial intelligence experts including Simon published the second volume of the Machine Learning collection, and the international magazine ML was founded, which further showed the rapid development trend of ML. The representative work of this stage includes Mostow’s guided learning, Lenat’s mathematical concept discovery program, Langley’s BACON program, and its improvement program.
Fourth Stage
The fourth stage, the mid-1980s, is the latest stage of machine learning. ML in this period has the following characteristics:
- Machine learning has become a new discipline, which comprehensively applies psychology, biology, neurophysiology, mathematics, automation, and computer science to form the theoretical basis of ML.
- A unified view of various basic issues of machine learning and artificial intelligence is taking shape.
- A variety of learning methods are integrated, and research on integrated learning systems in various forms is emerging.
- The application scope of various learning methods has been continuously expanded, and some applied research results have been transformed into products.
- Academic activities related to machine learning are unprecedentedly active.
Research Status
Machine learning is a day-to-day research hotspot in the field of artificial intelligence and pattern recognition, and its theories and procedure have been widely used to solve complex problems in engineering applications and science. The winner of the Turing Award in 2010 was Professor Leslie Valiant of Harvard University. One of his award-winning works was the establishment of the Probably Approximate Correct (PAC) learning theory; the winner of the Turing Award in 2011 was the University of California, Los Angeles. Professor Judea Pearl, whose main contribution is the establishment of artificial intelligence methods based on probability and statistics. These research results have contributed to the development and prosperity of machine learning.
Machine learning is a science that learning how to use computers to simulate or gain human learning activities. It is one of the most intelligent and cutting-edge research fields in AI(artificial intelligence). Since the 1980s, machine learning, as a way to realize artificial intelligence, has aroused widespread interest in the field of artificial intelligence. Mostly in the past 10 years, the research work in the area of ML has developed rapidly, and it has become the main part of artificial intelligence. one of the subjects.
Machine learning is used not only in knowledge-based systems but also in many fields such as natural language understanding, non-monotonic reasoning, machine vision, pattern recognition, etc. Whether a system has the ability to learn has become a sign of “intelligence”. The research of ML is mainly divided into two research directions: the first is the research of traditional ML, which mainly studies the learning mechanism, focusing on the exploration of the learning mechanism of simulated humans; the second is the research of machine learning in the environment of big data. Research, this type of research mainly studies how to effectively use information, focusing on obtaining hidden, effective, and understandable knowledge from huge amounts of data.
After 70 years of tortuous development, machine learning, represented by deep learning, draws on the multi-layered structure of the human brain, the layer-by-layer analysis, and processing mechanism of neuron connection and interaction information, and the powerful parallel information processing capability of self-adaptation and self-learning. There have been breakthroughs in the field of image recognition, the most representative of which is the field of image recognition.