Artificial intelligence is an emerging marginal discipline that develops in computer science, control, information theory, psychology, and language, etc., which focuses on the development of mutual penetration, mainly researching the use of machines (mainly computers) to imitate and achieve humanity. Intelligent behavior, after decades of development, artificial intelligence applications have developed in many areas, and there are also many places in our daily lives and learning. This article briefly introduces the application of symbol computing, pattern identification, expert system, machine translation, etc., which makes readers have a sense of understanding of artificial intelligence applications around us.
1 symbol calculation
One of the main purposes of the computer is scientific calculations, scientific calculations can be divided into two categories: a category of pure values, such as the value of the function, the numerical solution of the equation, such as weather forecast, reservoir simulation, aerospace, etc. . Another category is a symbol calculation, also known as an algebra, which is an intelligent calculation that is handled. Symbols can represent integers, yason, real numbers, and plural, or represent polynomial, function, collection, etc. . For a long time, people have always been looking forward to having a computer software system that can be symbolized. . As early as the end of the 1950s, people began to study. . After entering the 1980s, with the popularity of the computer and the development of artificial intelligence, there have been a variety of fully fully equipped computer algebraic system software, where Mathematica and Maple are their representatives, because they are written in C language, so you can Use on most computers. . Mathematica is the first mathematical software that combines symbolic operations, numerical calculations, and graphics, and users can easily use in a variety of forms of mathematical processing.
The superiority of the computer algebraic system is primarily that it can perform large-scale algebraic operations. . Usually, we use pen and paper to perform algebraic operations to handle less symbols. When the symbols of the formula rise to one hundred digits, manual calculations are very difficult, then use the computer algebraic system to make an accurate operation, Quick, effective. . Now there are some common feature of symbol computing software to be symbolic, numerical computing, and graphic display, etc., also have efficient programmable functions. In the operation interface, interactive processing is generally supported, and people are displayed by the keyboard input command. And the human-machine interface is friendly, the command is convenient and flexible, it is easy to seek help.
Although the computer algebra system has an extremely superiority in replacing the cumbersome symbolic operation, the computer is a machine after all, which can only perform the instructions to give it, there is a certain limit. First, most computer algebra systems have high requirements for computer hardware. When performing symbol operation, large memory and longer calculations are often required, while accurate algebra operations are at time and space. The second problem is to use the computer algebraic system to calculate, although the calculation accuracy can be used to any bit, but because the computer algebra system is the software itself, the software itself is used instead of hardware arithmetic operations, so in the speed ratio, the same problem is compared with the FORTRAN language. Slow hundreds of times or even thousands. In addition, although the computer algebraic system contains a large number of mathematical knowledge, this is just a small part of mathematics, and there are still many mathematics fields that have not been involved by the computer algebraic system. Computer algebra systems are still growing, perfect.
2 mode identification
Mode recognition is the automatic processing and judgment of the model by computer using mathematical methods. Here, we are collectively referred to as "model", and humans may study complex information processing processes with the development of computer technology. Automatic identification of computers (text, sound, characters, objects, etc.) is a key breakthrough in developing intelligent machines, and also provides clues for human awareness itself. An important form of the information processing process is the identification of life and object. For humans, it is particularly important to identify optical information (obtained by visual organs) and acoustic information (obtained by auditory official). This is two important aspects of pattern recognition. Representative products visible in the market have an optical character recognition, OCR, voice recognition system, and the like.
Computer identification is a high speed, accuracy and efficiency. The recognition process is similar to the learning process of human beings.
Take "Chinese Character Recognition" as an Example: First, the Chinese character image is processed, and the primary expression characteristics are extracted and the characteristics of the characteristics and the code of the Chinese characters are in the computer. Just like what to teach us this word, how to write knowledge memories in the brain. This process is called "training". The identification process is to compare the input Chinese character images after processing all the Chinese characters saved in the computer, identify the most similar words as identification results, this process is called "matching". Voice recognition is to let the computer can understand what people say, an important example is the seven-language language (English, day, hope, Korean, law, morality, middle) oral automatic translation system. Among them, the Chinese section of the experimental platform is set up in the model identification of the National Key Laboratory of Automation, Chinese Academy of Sciences, which is a sign of the world's leading level. After the system is implemented, people go abroad, buy tickets, when dialogue and exchange of foreign currency, as long as they use the telephone network and the Internet, mobile phones, calls, etc. can be called with "foreigners".
Fingerprint is an important feature of the human body, with uniqueness. The discrete geometry of the Beijing University conducted a deep study of the discrete geometry of the digital image, established the theoretical and algorithm of the precise calculation of the fingerprint grayscale image, and then extracted fingerprint feature information, and then successfully studied the identification of civil identification Automatic fingerprint appraisal system, and fingerprint appraisal system suitable for public security criminal detection. Thus, the primary river for the application of fingerprint automatic identification system in my country. The introduction of the automatic identification system of Peking University fingerprint has enabled my country's public security police to liberate the heavy manual treatment of fingerprint. Zhejiang Province has started using Peking University fingerprint automatic identification system, adopting the provincial (city) secondary construction library, provincial (city) county three-level query, forming unique "Zhejiang Mode". The Provincial Public Security Department has now established more than 1 million fingerprint libraries, and is the second largest library in China. In more than 1 million fingerprint libraries, only 4 minutes of the field fingerprint. In 2000, Zhejiang Province used fingerprint automatic recognition system to solve the case of 3063 cases, and there were more than 12,000 cases in conjunction. The case rate is the first in the country, and it is far ahead of the domestic fingerprint recognition system, and the public security department is used to apply models for fingerprint system.
Here is an example of a comprehensive application, and the FAW Group Corporation and the University of Defense Science and Technology have successfully developed the "Red Flag Car Independent Driving System" (ie unmanned driving system), which marks the ability of my country's development high-speed smart cars that has reached the world's advanced level. . Automotive autonomous driving technology is a comprehensive technology integrating multi-mate subjects such as pattern identification, intelligent control, computer learning and automotive manipulation, representing a national control technology. The red flag car self-driving system adopts computer visual navigation method, and uses imitation people to achieve manipulation control of the red flag. First, the camera inputs the road and the vehicle driving situation in front to the image processing and image recognition system. The system identifies road conditions, relative distances of the front vehicle and relative speed. Then, the path planning system plans a suitable path based on this information, which is determined how to drive. Then, the path tracking system combines the path to track, combined with the vehicle driving status parameters and vehicle driving dynamics constraints, forms a control command, controls the steering wheel, and the throttle opening mechanism to generate the corresponding action, so that the car is moving in accordance with the planned path, that is, drive according to the self-driving The system's planning path advances.
3 expert system
Expert system is a computer program system that simulates human experts to solve the problem of problems. The expert system contains a large number of expertise in a certain field, and can use the knowledge of human experts and solve problems, and simulate the decision-making process of human experts to solve complex problems in this area. Expert system is one of the most active and broadest applications of artificial intelligence applications. It involves all aspects of society. Various expert systems have spread all major, and they have achieved great success. According to the type of expert system processing, the expert system is divided into 10 types of interpretation, diagnostic, commissioning, maintenance, education, predictive, planning, design and control type. Specific applications are much more, such as blood condensed disease diagnostic systems, telephone cable maintenance expert system, flower cloth pattern design and flower printing and dyeing expert system, etc.
In order to achieve an expert system, it is necessary to store the expertise in the special field, analyzed and analyzed and analyzed according to some model (composition knowledge base), and a reasoning mechanism that is similar to the domain expert to solve practical problems (constitutes a throttle) . The system can process the input information, and use knowledge to reason to reason, and make decisions and judgments. The level of solving problems reaches or approaches the expert's level, so it can play an expert or expert assistant. The key to the development expert system is to express and use expert knowledge, that is, the facts and processes that have been proven to address typical issues in the relevant fields. At present, the expert system mainly adopts rule-based knowledge representation and reasoning technology. Due to more knowledge in the field is not accurate or uncertain, uncertain knowledge representation and knowledge reasoning is an important topic of expert system development and research. In addition, the development and development of expert system development tools is also very rapid. This will actively promote the development process of expanding the application of expert systems and accelerating expert systems. With the overall level of computer science and technology, the research of new generation experts such as distributed expert systems, collaborative expert systems, has also developed very quickly. In a new generation expert system, it is not only used in rule-based reasoning methods, but also methods and techniques such as artificial neural networks.
4 machine translation
The machine translation is the process of transfining a natural language into another natural language to complete the software system of this process is called machine translation system. For decades, many experts at home and abroad have paid a lot of effort and sweat for machine translation. Although there is no practical, comprehensive, high-quality automatic translation system, but also has made great progress, especially as people's auxiliary translation tools, machine translation has been recognized by most people. At present, the domestic machine translation software is not 100 kinds of translation characteristics according to these software, which can be roughly divided into three categories: dictionary translation, Chinese translation class and professional translation. The translation software of the dictionary is "Jinshan Words", which is called how fast and good electronic dictionary, it can quickly query the word meaning of English words or phrase, and provide words of pronunciation, providing users to provide a very word or phrase meaning Large convenience. The typical representative of Chinese translation software is "Oriental Express 2000", which first proposes the concept of "smart Chinese", making the auxiliary translation of translation software more obvious. The professional translation system represented by "Translation Star" and "Ya Xun translation" is the translation software for professional or industry users, but the quality of its professional translation has many gaps, some people commented " Full of English can't live, full of Chinese can't understand ", the saying is extremely extreme, but the quality of the translation translation is indeed a difficult problem. Here, we may wish to make a comparison of existing machine translation and interpretation, and we can see some reasons.
machine translation:
1. One sentence is handled, the context lacks contact;
2. The analysis of the source language is just to solve the syndrome relationship, which is not a sense of understanding;
3. Lack of domain knowledge, from computer to medicine, from chemical industry to legal, change professional dictionary;
4. Translation conversion is the constraints of the syntactic structure of the source language, the syntact structure of the source language;
5. Translation is just a mechanical corresponding to the structure of the syntax.
Artificial translation:
1. Generally read through the full text, and should be taken before and after;
2. The source language is the understanding of the meaning;
3. Only professional translators, without universal translators;
4. The translation is based on his understanding of the source language, and the constraints of the syntactic structure of the source language;
5. Translation is a re-created process.
In the current situation, computer-assisted translations should be a relatively good practical choice. In fact, in many fields, computer-assisted humans have been widely used, such as CAD software. If computer-aided technology is used in language translation research, it should also be a large auxiliary role, which is called "computer-aided translation". It integrates the machine memory translation, grammatical analysis and interpersonal translation, and handles mechanical, repetition, trivial work in the translation process to the computer. In this way, the translator only needs to concentrate on creative thinking, which is conducive to the improvement of productivity. The machine translation study is a knowledge processing problem that involves knowledge, language inter-language knowledge, and world knowledge outside the language, including common knowledge and related fields. With the popularity and development of the Internet, the application prospects for machine translation is very broad. As a window that explores the mechanism of intelligence and operation knowledge, the research and application of machine translation will be more attractive. International analysis of experts believed that machine translation should achieve the same smoothness like artificial translation, at least 15 years of continuous research, but in terms of mankind, there is no clear "how the human brain is ambiguous" In the case of judging, it is impossible to reach 100% accuracy.
5 Application of Artificial Intelligence Thoughts: Looking for Aliens at Home
The basic idea of artificial intelligence has been applied in many fields, "Searching at home" (SETI @ Home) project is a successful case with artificial intelligence neural network and grid calculation ideas. SETI @ Home is the abbreviation of Search for extra Terrestrial Intelligence At Home, which is: Looking for alien civilization at home. The project was launched by the US Planet Society and the University of California, and the US California University started from May 17, 1999. The data obtained helps scientists explore alien creatures. The essence of its computing mode is the grid calculation.
The approximate process of the seti @ Home project is this:
1. Government or research departments submit a task that requires a huge amount of operations to the server in the form of procedures and data.
2. The server divides the data and program code into smaller parts, also known as "sub-task".
3. Install a special customer program on the volunteer's PC (in fact a screen saver), it can automatically contact the server, automatically download and process subtutaneous tasks.
4. The result is sent back to the server after the sub task is processed. Then, the client program downloads a new submission to continue processing.
5. Once all sub tasks are processed, the server summarizes the various results, generates the final report, and send the final result back to the author.
Seti @ Home's information is derived from the information collected by the Radio telescope of Puendo Rico, this observation begins with approximately 35g of information every day, because there is not enough online bandwidth, so it can only be recorded Tape, served in Berklai in a traditional mailing manner, and the data is divided into approximately 350K per unit, transmitting the participant to the participant through the Internet, with a dispersion operation to handle huge data. Since the SETI @ Home project did not get a special funding fee of the country, I have to ask the PC to help around the world. People who participated in this event did not have any rewards, only silently dedication, but how much statistics were found according to feedback, although the aliens did not find so far, the SETI @ Home project strongly proved that the grid calculation is Effective.
Since the formal launch of the SETI @ Home project, there have been some pieces of information collected by the information collected from the specified site, and use their computer to run analysis, from which to find signs of life in the universe. The total processing data reached 15T, and the average of each participant made his computer for 17 and a half hours, which is equivalent to using a PC working for 48 2023, which is equivalent to using supercomputer working for 48 years. The project fully utilizes the power distributed around the world, although only $ 500,000, it has a powerful power. Interested readers can log in
Http://setiathome.ssl.berkeley.edu/home_chinese.html Website Learn or personally participate in the project.
