Artificial Intelce
人工智能
Advanced Idea, Anticipating Incomparability[1]—on AI, Artificial Intelce
Artificial intelce (AI) is the field of engineering that builds systems, primarily computer systems, to perform tasks requiring intelce. This field of research has often set itself ambitious goals,seeking to build machines that can "outlook" humans in particular domains of skill and knowledge, and has achieved some success in this aspect. The key aspects of intelce around which AI research is usually focused include expert system[2], industrial robotics, systems and languages, language understanding, learning, and playing, etc.
Expert System
An expert system is a set of programs that manipulate encoded knowledge to solve problems in a specialized domain that normally requires human expertise. Typically, the user interacts with an expert system in a "consultation dialogue", just as he would interact with a human who had some type of expertise—explaining his problem,performing suggested tests, and asking questions about proposed solutions. Current experimental systems have achieved high levels of performance in consultation tasks like chemical and geological data ysis, computer system configuration, structural engineering, and even medical diagnosis. Expert systems can be viewed as intermediaries between human experts, who interact with the systems in "knowledge acquisition" mode[3], and human users who interact with the systems in "consultation mode". Furthermore, much research in this area of AI has focused on endowing these systems with the ability to explain their reasoning, both to make the consultation more acceptable to the user and to help the human expert find errors in the system's reasoning when they occur. Here are the features of expert systems.
① Expert systems use knowledge rather than data to control the solution process.
② The knowledge is encoded and maintained as an entity[4] separated from the control program. Furthermore, it is possible in some cases to use different knowledge bases with the same control programs to produce different types of expert systems. Such systems are known as expert system shells[5].
③ Expert systems are capable of explaining how a particular conclusion is reached, and why requested information is needed during a consultation.
④ Expert systems use symbolic representations for knowledge and perform their inference through symbolic computations[6].
⑤ Expert systems often reason with metaknowledge.
Industrial Robotics
An industrial robot is a general-purpose computer-controlled manipulator consisting of several rigid links connected in series by revolute or prismatic joints[7]. Research in this field has looked at everything from the optimal movement of robot arms to methods of planning a sequence of actions to achieve a robot's goals. Although more complex systems have been built, thousands of robots that are being used today in industrial applications are devices that have been programmed to perform some repetitive tasks. Robots, when compared to humans, yield more consistent quality, more predictable output, and are more reliable. Robots have been used in industry since 1965. They are usually characterized by the design of the mechanical system. There are six recognizable robot configurations:
① Cartesian Robots[8]: A robot whose main frame consists of three linear axes[9].
② Gantry Robots[10]: A gantry robot is a type of artesian robot whose structure resembles a gantry. This structure is used to minimize deflection along each axis.
③ Cylindrical Robots[11]: A cylindrical robot has two linear axes and one rotary axis.
④ Spherical Robots[12]: A spherical robot has one linear axis and two rotary axes. Spherical robots are used in a variety of industrial tasks such as welding and material handling.
⑤ Articulated Robots[13]: An articulated robot has three rotational axes connecting three rigid links and a base.
⑥ Scara Robots: One style of robot that has recently become quite popular is a combination of the articulated arm and the cylindrical robot. The robot has more than three axes and is widely used in electronic assembly.
Systems and Languages
Computer-systems ideas like timesharing, list processing, and interactive debugging were developed in the AI research environment[14]. Specialized programming languages and systems, with features designed to facilitate deduction, robot manipulation, cognitive modeling, and so on, have often been rich sources of new ideas. Most recently, several knowledge-representation languages—computer languages for encoding knowledge and reasoning methods as data structures and procedures—have been developed in the last few years to explore a variety of ideas about how to build reasoning programs.
Problem Solving
The first big "success" in AI was programs that could solve puzzles and play s like chess. Techniques like looking ahead several moves and dividing difficult problems into easier sub-problems evolved into the fundamental AI techniques of search and problem reduction. Today's programs can play championship-level checkers and backgammon, as well as very good chess. Another problem-solving program that integrates mathematical formulates symbolically has attained very high levels of performance and is being used by scientists and engineers.Some programs can even improve their performance with experience.
As discussed above, the open questions in this area involve capabilities that human players have but cannot articulate, like the chess 's ability to see the board configuration in terms of meaningful patterns. Another basic open question involves the original conceptualization of a problem, called in AI the choice of problem representation. Humans often solve a problem by finding a way of thinking about it that makes the solution easy—AI programs, so far, must be told how to think about the problems they solve.
Logical Reasoning
Closely related to problem and puzzle solving was early work on logical deduction[15]. Programs were developed that could "prove" assertions by manipulating a database of facts, each represented by discrete data structures just as they are represented by discrete formulas in mathematical logic. These methods, unlike many other AI techniques, could be shown to be complete and consistent. That is, so long as the original facts were correct, the programs could prove all theorems that followed from the facts, and only those theorems.
Logical reasoning has been one of the most persistently investigated subareas of AI research. Of particular interest are the problems of finding ways of focusing on only the relevant facts of a large database and of keeping track of the justifications for beliefs and updating them when new information arrives.
Language Understanding
The domain of language understanding was also investigated by early AI researchers and has consistently attracted interest. Programs have been written that answer questions posed in English from an internal database, that translate sentences from one language to another, that follow instruction given in English, and that acquire knowledge by reading textual material and building an internal database. Some programs have even achieved limited success in interpreting instructions spoken into a microphone instead of typed into the computer. Although these language systems are not nearly as good as people are at any of these tasks, they are adequate for some applications. Early successes with programs that answered queries and followed directions, and early failures at machine translation, have resulted in a sweeping change in the whole AI approach to language. The principal themes of current language-understanding research are the importance of vast amounts of general, commonsense world knowledge and the role of expectations, based on the subject matter and the conversational situation, in interpreting sentences.
Learning
Learning has remained a challenging area for AI. Certainly one of the most salient and significant aspects of human intelce is the ability to learn. This is a good example of cognitive behavior that is so poorly understood that very little progress has been made in achieving it in AI systems[16]. There have been several interesting attempts, including programs that learn from examples, from their own performance, and from being told. An expert system may perform extensive and costly computations to solve a problem. Most expert systems are hindered by the inflexibility of their problem-solving strategies and the difficulty of modifying large amounts of code. The obvious solution to these problems is for programs to learn on their own, either from experience, ogy, and examples or by being "told" what to do.
Playing
Much of the early research in state space search was done using common board s such as checkers, chess, and the 15-puzzle. In addition to their inherent intellectual appeal, board s have certain properties that make them ideal subjects for this early work. Most s are played using a well-defined set of rules, which makes it easy to generate the search space and frees the researcher from many of the ambiguities and complexities inherent in less structured problems. The board configurations used in playing these s are easily represented on a computer, requiring none of the complex formalisms.
Conclusion
We have attempted to define artificial intelce through discussion of its major areas of research and application. In spite of the variety of problems addressed in artificial intelce research[17], a number of important features emerge that seem common to all divisions of the field, including.
① The use of computers to do reasoning, learning, or some other forms of inference.
② A focus on problems that do not respond to algorithmic solutions. This underlies the reliance on heuristic search[18] as an AI problem-solving technique.
③ Reasoning about the significant qualitative features of a situation.
④ An attempt to deal with issues of semantic meaning[19] as well as syntactic form[20].
⑤ The use of large amounts of domain-specific knowledge in solving problems. This is the basis of expert systems.
Notes
[1] 标题中的两个短语分别为两组AI，以此分别强调人工智能的最新理念无与伦比。
[2] expert system专家系统。
[3] "knowledge acquisition" mode知识获取模式。
[4] entity实体。
[5] expert system shells专家系统外壳。
[6] symloolic computation符号计算。
[7] ...by revolute or prismatic joints通过外卷的，或棱镜似的连接结合起来。
[8] Cartesian Robot直角座标机器人，主框架由三根直线轴构成。
[9] linear axes线性轴。
[10] Gantry Robot桶架式机器人Gantry桶架。
[11] Cylindrical Robot or Cylindrical Coordinate Robot柱面坐标式机器人。
[12] Spherical Robot or Spherical Coordinate Robot球坐标式机器人。
[13] Articulated Robot挂接式机器人。
[14] Computer-systems ideas like time-sharing, list processing, and interactive debugging were developed in the AI research environment. 人工智能采用了计算机系统方面的一些理念，如：时间分配，编目处理，交互式调试，等等。
[15] logical deduction逻辑推断(演绎推理的过程，在此过程中必然可从所述前提得出一个结论；从一般推向特殊的推论)。
[16] This is a good example of cognitive behavior that is so poorly understood that very little progress has been made in achieving it in AI systems. 这是一种典型的认知行为，但人们却不太了解它，以至于人工智能在这方面还没有什么发展。
[17] In spite of the variety of problems addressed in artificial intelce research. 尽管人工智能研究中出现了各种各样的问题……
[18] heuristic search启发式搜索。
[19] semantic meaning语义(计算机语言中的每个语义成分所代表的实际操作)。
[20] syntactic form语法形式；句法形式。