The main contents of artificial intelligence research include: knowledge representation, automatic reasoning and search methods, machine learning and knowledge acquisition, knowledge processing system, natural language understanding, computer vision, intelligent robots, automatic programming and so on.
Knowledge representation is one of the basic problems of artificial intelligence, and reasoning and searching are closely related to representation methods. Commonly used knowledge representation methods include: logical representation, production representation, semantic network representation and frame representation.
Common sense naturally attracts people's attention, and people have put forward various methods such as non-monotonic reasoning and qualitative reasoning to express and deal with common sense from different angles.
Automatic reasoning in solving problems is the process of applying knowledge. Because there are many knowledge representation methods, there are also many corresponding reasoning methods. The reasoning process can generally be divided into deductive reasoning and non-deductive reasoning. Predicate logic is the basis of deductive reasoning. Inheritance performance reasoning under structured representation is non-deductive. Due to the need of knowledge processing, in recent years, people have proposed a variety of non-deductive reasoning methods, such as connection mechanism reasoning, analogy reasoning, case-based reasoning, deductive reasoning and restrictive reasoning.
Search is a problem solving method of artificial intelligence, and the search strategy determines the priority of knowledge used in a reasoning step of problem solving. It can be divided into blind search without information guidance and heuristic search guided by empirical knowledge. Heuristic knowledge is usually represented by heuristic functions. The more fully heuristic knowledge is used, the smaller the search space for solving problems. Typical heuristic search methods include A*, AO* algorithm and so on. In recent years, the research of search methods has begun to pay attention to the ultra-large-scale search problem of millions of nodes.
Machine learning is another important topic of artificial intelligence. Machine learning refers to the process of acquiring new knowledge in a certain sense of knowledge representation. According to the different learning mechanisms, there are mainly inductive learning, analytical learning, linkage mechanism learning and genetic learning.
Knowledge processing system is mainly composed of knowledge base and inference engine. Knowledge base stores the knowledge needed by the system. When the amount of knowledge is large and there are many representations, it is very important to organize and manage the knowledge reasonably. The inference engine specifies the basic methods and strategies of applying knowledge when solving problems. In the process of reasoning, it is necessary to establish a database or use blackboard mechanism to record the results or communicate. If the expert knowledge in a certain field (such as medical diagnosis) is stored in the knowledge base, such a knowledge system is called an expert system. In order to meet the needs of solving complex problems, a single expert system is developing into a multi-agent distributed artificial intelligence system. At this time, knowledge sharing, cooperation among subjects and the generation and handling of contradictions will be the key issues of research.
First, the history of artificial intelligence
Artificial intelligence (AI) is a challenging science, and people engaged in this work must understand computer knowledge, psychology and philosophy. Artificial intelligence includes a wide range of sciences and consists of different fields, such as machine learning, computer vision and so on. Generally speaking, the purpose of artificial intelligence is to make computers think like people. This is not an easy thing. If you want to build a thinking machine, you must know what thinking is, and further, what wisdom is and what its performance is. You can say science.
Home is wise, but you will never say that a passerby knows nothing and has no knowledge. You dare not say that children are not intelligent, but you dare not say that they are intelligent for a machine. So how to distinguish wisdom? What we say, what we do, our thoughts flow out of our brains like spring water, so natural, but machines can, so what kind of machines are intelligent? Scientists have made cars, trains, planes, radios and so on. They imitate the functions of our body organs, but can they imitate the functions of the human brain? So far, we only know that this thing in our crown is an organ composed of billions of nerve cells. We know very little about this thing, and imitating it may be the most difficult thing in the world.
British scientist Turing contributed to the definition of wisdom. A machine is intelligent if it can pass an experiment called Turing Experiment. The essence of Turing experiment is that when people can't distinguish the behavior of machines from that of people without looking at their appearance, machines are intelligent. Don't think that Turing will go down in history only after making this contribution. If you study computers, you will know that for computer people, winning the Turing Prize is equivalent to winning the Nobel Prize for physicists. Turing laid the foundation for the emergence of computers in theory. Without his outstanding contribution, there would be no such thing in the world, let alone the internet.
Long before the advent of computers, scientists hoped to create machines that could simulate human thinking. In this regard, I would like to mention another outstanding mathematician and philosopher, Boolean. Together with other outstanding scientists, he laid the thinking structure and method of intelligent machines through the mathematical and accurate description of human thinking, and the logical foundation used by our computers today was founded by him.
I think anyone who has studied computer will be familiar with Boolean. It created the Boolean algebra we learned. When the computer appeared, human beings began to really have a tool that can simulate human thinking. In the following years, countless scientists worked hard for this goal. Now artificial intelligence is not the patent of several scientists. Almost all computer departments of universities in the world are studying this subject, and college students who study computer must also study such a course. Through unremitting efforts, computers now seem to have become very smart. It is well known that computers beat people in the just-concluded chess match. You may not have noticed that in some places, computers help people do other jobs that originally belonged to human beings. Computers play a role for human beings with their high speed and accuracy. Artificial intelligence has always been the frontier subject of computer science, and computer programming languages and other computer software also exist because of the progress of artificial intelligence.
Now human beings have improved the computing power of computers to an unprecedented level, and artificial intelligence will also lead the trend of computer development in the next century. Now the development of artificial intelligence is not obvious because of theoretical limitations, but it will certainly affect our lives as profoundly as today's network.
The research on artificial intelligence has been started for a long time all over the world, but the real realization of artificial intelligence should be counted from the birth of computers, and then it is possible for human beings to realize human intelligence with machines. The English word AI was first put forward at a meeting in 1956. After that, it developed with some scientific efforts. The progress of artificial intelligence is not as rapid as we thought, because the basic theory of artificial intelligence is not complete, and we can't essentially explain why our brain can think, what this thinking comes from, why this thinking comes into being, and so on. But after decades of development, artificial intelligence is affecting people's lives with its great power.
Let's review the development of computers with the development of artificial intelligence. 194 1 year, the first computer jointly developed by the United States and Germany was born. Since then, the method of storing and processing information has undergone revolutionary changes. The appearance of the first computer was not very good. It is fat and delicate, and needs to work in an air-conditioned room. If you want it to handle anything, you need to reconnect the wires. This is not a labor-saving job. Thousands of silk. I think programmers are living in heaven now.
Finally, a computer that can store programs was invented in 1949. So that the programming program can finally be soldered, which is much better. Because programming becomes very simple, the development of computer theory finally leads to the emergence of artificial intelligence theory. People can finally find a way to store information and process it automatically.
Although it seems that this new machine can realize some human intelligence now, it was not until the 1950s that people linked human intelligence with this new machine. We noticed the old man with a big belly next to him. His research on feedback theory finally made him put forward an argument. All this
The results of human intelligence are all the results of feedback, and intelligence is produced by constantly feeding back the results to the body. Our toilet is a good example. The reason why water does not stop flowing is precisely because there is a device to detect the change of water level. If there is too much water, turn off the water pipe, which realizes feedback and is a negative feedback. If even the equipment in our toilet can achieve feedback, then we should be able to achieve feedback with a machine, and then realize the machine form replication of human intelligence. This idea had a great influence on the early days of artificial intelligence.
At the time of 1955, Shannon and others developed the logic theorist program, which is a program with a tree structure. When the program runs, it searches the tree and finds the branch of the tree closest to the possible answer to get the correct answer. This program can be said to have an important position in the history of artificial intelligence, which has brought great influence to the academic community and society, so that many methods and ideas we use now still come from this program in the 1950s.
1956, McCarthy (the man on the right), another famous scientist in the field of artificial intelligence, convened a meeting to discuss the future development direction of artificial intelligence. Since then, the name of artificial intelligence has been formally established. This conference was not a great success in the history of artificial intelligence, but it gave the founders of artificial intelligence an opportunity to communicate with each other and paved the way for the future development of artificial intelligence. After that, the focus of workers' intelligence is to establish a practical system that can solve problems by itself, and requires the system to have self-learning ability. 1957, Shannon and others developed a program called General Problem Solver (GPS), which extended Wiener's feedback theory and could solve some common problems. While other scientists are trying to develop this system, the scientist on the right has made great contributions. He created the form processing language LISP, which is still used by many artificial intelligence programs, and it has almost become synonymous with artificial intelligence. Today, LISP is still developing.
1963, MIT is supported by the U.S. government and the Ministry of National Defense to carry out artificial intelligence research. The American government did nothing but keep a balance with the Soviet Union during the Cold War. Although this purpose is a bit explosive, its result has greatly developed artificial intelligence. Since then, many programs have attracted much attention, and MIT has developed SHRDLU. In 1960s, the student system could solve algebra problems, while the SIR system began to understand simple English sentences. The emergence of SIR led to the emergence of a new discipline: natural language processing. The emergence of expert system in 1970s is a great progress. People know for the first time that computers can replace human experts. Due to the improvement of computer hardware performance, artificial intelligence can carry out a series of important activities, such as statistical analysis of data and participation in medical diagnosis. As an important aspect of life, it began to change human life. Theoretically, the 1970s was also a period of great development, and computers began to have simple thinking and vision. However, in the 1970s, another artificial intelligence language, Prolog, was born. Together with LISP, it almost became an indispensable tool for artificial intelligence workers. Don't think that artificial intelligence is far away from us. It has entered our life, fuzzy control, decision support and so on. It is the purpose of artificial intelligence to let computers take the place of human beings to carry out simple intellectual activities and liberate human beings to engage in other more beneficial work, but I think the endless pursuit of scientific truth is the ultimate driving force.
Second, the application field of artificial intelligence
1, solve the problem.
The first great achievement of artificial intelligence is the chess program. Some techniques used in chess games, such as looking forward a few steps, break down difficult problems into some easier sub-problems and develop them into basic artificial intelligence techniques, such as searching and problem induction. Today's computer programs have reached the champion level of various squares and chess. But it has not been solved, including the ability that human players have but cannot express clearly. For example, the chess master's insight into the chess game. Another problem is related to the original concept of the problem, which is called the choice of problem representation in artificial intelligence. People can often find some ways to think about problems, thus making solutions easier and solving problems. So far, artificial intelligence programs have been able to know how to consider the problem they want to solve, that is, search the solution space and find a better solution.
2. Logical reasoning and theorem proving.
Logical reasoning is one of the most persistent fields in artificial intelligence research. It is particularly important to find some methods that only focus on the relevant facts in large databases, pay attention to credible proofs, and correct these proofs in time when new information appears. Conjecture in mathematics. Finding a theorem to prove or disprove requires not only the ability to deduce according to assumptions, but also many informal tasks, including medical diagnosis and information retrieval, can be formalized like theorem proving. Therefore, theorem proving is an extremely important topic in the research of artificial intelligence methods.
3. Natural language processing.
Natural language processing is a typical example of the application of artificial intelligence technology in practical fields. After years of hard work, many remarkable achievements have been made in this field. At present, the main topic in this field is: how to generate and understand natural language based on themes and dialogue situations, paying attention to a lot of common sense-world knowledge and expectations. This is an extremely complicated coding and decoding problem.
4. Intelligent information retrieval technology.
Influenced by the rapid development of "() *+(*)" technology, information acquisition and refining technology has become an urgent research topic in contemporary computer science and technology research. The application of artificial intelligence technology in this field is an opportunity and breakthrough for the wide application of artificial intelligence in practice.
5. Expert system.
Expert system is the most active and effective research field in artificial intelligence at present. It is a program system with a lot of knowledge and experience in a specific field. In recent years, there has been a trend of successful and effective application of artificial intelligence technology in the research of "expert system" or "knowledge engineering". Human experts have rich knowledge, so they can achieve excellent problem-solving ability. So if the computer program can embody and apply this knowledge, it should also be able to solve the problems solved by human experts and help human experts find the mistakes in the reasoning process, which has been confirmed now. For example, in mineral exploration, chemical analysis, planning and medical diagnosis, expert systems have reached the level of human experts. A successful example is that the prospecting system found a molybdenum deposit worth more than 654.38 billion dollars. The performance of DENDRL system has exceeded the level of ordinary experts, and it can be used by hundreds of people in chemical structure analysis. My CIN system can provide suggestions for the diagnosis and treatment of blood infectious diseases. After formal identification, the diagnosis and treatment of bacterial hematological diseases and meningitis have surpassed experts in this field.