* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Artificial Intelligence
Survey
Document related concepts
Machine learning wikipedia , lookup
Computer vision wikipedia , lookup
Human–computer interaction wikipedia , lookup
Artificial intelligence in video games wikipedia , lookup
Technological singularity wikipedia , lookup
Computer Go wikipedia , lookup
Embodied cognitive science wikipedia , lookup
Knowledge representation and reasoning wikipedia , lookup
Intelligence explosion wikipedia , lookup
Philosophy of artificial intelligence wikipedia , lookup
Ethics of artificial intelligence wikipedia , lookup
Existential risk from artificial general intelligence wikipedia , lookup
Transcript
Artificial Intelligence ARTIFICIAL INTELLIGENCE Venkata Phani Anne IMSC500-1902-Information System Technology Professor Soroushi The University of Northern Virginia March 18, 2010 Page 1 Artificial Intelligence ABSTRACT This paper describes the development of an application of Artificial Intelligence (AI) in game playing, speech appreciation, understanding natural language and Computer version. And describes the solution of specific problems, longstanding differences of opinion about how AI should be done and the applications of widely differing tools. The most problems of AI include lot of verities as Communication, development, learning, understanding, ability to move and acuity. Page 2 Artificial Intelligence CONTENENTS Sl. No. 1.0 2.0 3.0 3.1 3.2 3.3 3.4 3.5 3.6 3.7 4.0 4.1 4.2 4.3 4.4 4.5 5.0 5.1 5.2 5.3 5.4 5.5 6.0 6.1 6.2 6.3 6.4 6.5 6.6 7.0 8.0 Description Description History Problems Learning Motion and manipulation Deduction Reasoning General intelligence Planning Social intelligence Approaches Integrating approaches Brain simulation cognitive architectures Knowledge based Symbolic Tools Probabilistic method Search and optimization Logic Classifiers and statistical learning methods Evaluating progress Applications Game playing Speech recognition Understanding natural language Computer vision Expert systems Heuristic classification Uses Future References Page No. 4 4 5 5 6 6 6 6 7 7 7 7 8 8 8 9 9 9 10 11 12 12 13 13 14 14 14 14 15 15 16 18 Page 3 Artificial Intelligence 1.0 DEFINITION:Artificial intelligence is a field of science and engineering concerned with the computational perceptive of what is commonly called intelligent behavior, and with the creation of artifacts that exhibit such behavior. 2.0 HISTORY:The history of AI is a history of fantasies, prospective, demonstrations, and guarantee. Ever since Homer wrote of mechanical “tripods” waiting on the gods at dinner, imagined mechanical assistants have been a part of our traditions. However, only in the last half century have we, the AI community, been able to build untried machines that test hypotheses about the mechanisms of thought and intelligent behavior and thereby display mechanisms of thought and intelligent behavior and there by demonstrate mechanisms that before existed only as theoretical potential. Although achieving full-blown artificial intelligence remains in the ongoing conversation about the implications of realizing the promise. Philosophers have floated the opportunity of intelligent machines as a fictitious device to help us define what it means to be human. Rene Descartes, for example, seems to have been more interested in “mechanical man” as a metaphor than as a possibility. Gottfried Wilhelm Leibniz, on the other hand, seemed to see the possibility of mechanical reasoning devices using rules of logic to settle disputes. Both Leibniz and Braise Pascal designed calculating machines that mechanized arithmetic, which had hitherto been the area, learned men called “calculators,” but they never made the claim that the devices could think. Etienne Bonnet, Abbey de Cadillac used the metaphor of a statue in to whose head we poured nuggets of knowledge, asking at what point it would know enough to appear to be intelligent. Page 4 Artificial Intelligence Robots and artificially shaped beings such as the Golem in Jewish tradition and Mary Shelly’s Frankenstein have always captured the public’s imagination, in part by playing on our fears. Mechanical animals and dolls-including a mechanical trumpeter for which Ludwig van Beethoven wrote a trumpet blast- were actually built from clockwork mechanisms in the seventeenth century. Although they were apparently limited in their performance and were intended more as curiosities than as demonstrations of thinking, they provided some initial credibility to mechanistic views of actions and to the idea that such actions need not be feared. As the industrial world became more mechanized, machinery became classier and more humdrum. But it was still essentially clockwork. Chess is quite obviously an enterprise that requires thought. It is not too surprising, then, that chess-playing machines of the eighteenth and nineteenth centuries, most notably “the Turk,” were exhibited as intelligent machines and even fooled some people into believing the machines were playing by you. Chess was widely used as a vehicle for studying supposition and depiction mechanisms in the early decades of AI work. 3.0 Problems: The general problem has been broken down in to a number of sub-problems. The following are the different problems 3.1 Learning: Natural language processing gives machines the ability to read and understand the language that humans speak. Unsupervised learning is the ability to find patterns in a stream of input. Supervised learning includes both classification and numerical regression. Classification is used to determine what category something belongs in, after seeing a number of examples of things several categories. Regression takes a set of numerical output/input. Page 5 Artificial Intelligence 3.2 Motion and manipulation: Robotics field is related to AI. Intelligence is required for robots to be able to grip such errands as object management and steering. 3.3 Deduction: Human beings determine most of their problems using fast, impulsive judgments rather than the conscious, step-by-step assumption that early was able to model. For difficult problems, most of these algorithms can need huge computational funds the amount of memory or computer time requires becomes sky-high when the problem goes beyond a positive size. 3.4 Reasoning: AI has made some progress at imitating this kind of “sub-symbolic” problem solving. alive agent approaches lay emphasis on the importance sensor motor skills to higher reasoning. 3.5 General intelligence: Many of the problems above are considered complete to solve one problem, you must solve them all. Combining all the skills above and exceeding human abilities at most or all of them, a few believe that anthropomorphic features like consciousness are an artificial brain may be required for such a project. Even a basic, specific task like machine translation requires that the machine follow the author’s argument, know what is being talked about and faithfully reproduce the attention. Page 6 Artificial Intelligence 3.6 Planning: They need away to dream of the future and be able to make choices that maximize the effectiveness of the available choices. Planning must be capable to set goals and achieve them. In classical problem solving, the agent can assume that it is the only thing acting on the world and it can be certain what the penalty of it action may be. 3.7 Social intelligence: Two roles of an intelligent agent are emotion and social skills play. a) It must be able to predict the actions of others, by understanding their motives and emotional stats. b) For good human-computer communication, an intelligent machine also needs to display emotions. At very least it must appear polite and sensitive to the human it interacts with. 4.0 Approaches; A few of the most long standing questions that have remained unanswered should AI simulate neutral intelligence, by studying psychology biology as irrelevant to AI research as bird biology . Can intelligent behavior to described using simple, elegant principles. 4.1 Integrating the approaches An agent that solves a specific problem can use any approach that works – some agents are symbolic and logical, some are sub-symbolic neutral networks and other may use new approaches. Intelligent agent paradigm is a system that perceives its environment and takes actions which maximizes its chances of success. Page 7 Artificial Intelligence 4.2 Brain simulation: A Manu of the researchers gathered for meetings to the teleological society approach was largely abandoned, although elements of it would be revived. A system with both symbolic and sub symbolic components is a hybrid intelligent system, and the study of such system is artificial intelligence systems integration. 4.3 cognitive architectures: The connection between the neurology, information theory, and cybernetics, some of them built machines that used electronic networks to exhibit simple intelligence. A hierarchical control system provides a bridge between sub symbolic AI at its lowest, reactive levels and traditional symbolic AI at its highest levels, where relaxed time constraint permit and world modeling. 4.4 Knowledge based: The knowledge evolution led to the development and consumption of expert systems the first truly successful form of Ai software. When computers with large memories became available around 1970, researchers from all three traditions began to build knowledge AI applications. The knowledge revolution was also driven by the consciousness that enormous amounts of knowledge would be required by many simple AI applications Page 8 Artificial Intelligence 4.5 Symbolic: Logic based, machines did not need to simulate human thought, but should instead try to find essence of abstract reasoning and problem solving, regardless of whether people used the same algorithms. Cosngnitive simulation studied human problem solving skills and attempted to finalize them, and their work laid the foundations of the field of artificial intelligence, as well as cognitive science, operations research and management science. 5.0 Tools: Tools are most important once to solve the most difficult problems AI has developed large number of tools. The following are the tools mostly used in 5.1 Probabilistic method for uncertain reasoning Bayesian networks are a very general tool that can be used for large number of problems reasoning, learning, planning and perception. Many problems in AI in reasoning, planning, learning, perception and robotics require the agent to operate with incomplete or uncertain information. Starting in the late 80s and early 90s, Judea pearl and others championed the use of methods drawn from probability theory and economics to devise a number of powerful tools to solve these problems. Probabilistic algorithms can also be used for filtering, prediction, smoothing and finding explanations for streams of data, helping perception systems to analyze process that occur over time. Page 9 Artificial Intelligence A key concept from the science of economics is utility a measure how valuable something to an intelligent agent. Precise mathematical tools have been developed that analyze how dynamic decision networks, game theory and mechanism design. 5.2 Search and optimization They may begin with a population of organisms and then allow them to mutate and recombine, selecting only the fittest to survive each generation. Many problems in AI can be solved in theory by intelligently searching through many possible solutions. A very different kind of search came to prominence in the 1990s, based on the mathematical theory of optimization. Planning algorithms search through trees of goals and sub goals, attempting to find a path to a target goal, a process called means ends analysis. Reasoning can be reduced to performing a search.Logical proof can be viewed as searching for a path that leads from premises to conclusions, where each step is the application of an inference rule. Robotics algorithms for moving limbs and grasping object use local searches in configuration space. Many learning algorithms use algorithms based on optimization. Simple exhaustive searches are rarely sufficient for most real world problems. The searches grow quickly to astronomical numbers. These algorithms can be visualized as blind hill climbing. We begin the search t a random point on the landscape, and then, by jumps or steps, we moving our guess uphill, until we reach the top. The result is a search that is too slow or never completes. The solution, for many problems, is to use rules of thumb that eliminate choices that are unlikely to lead to the goal. Heuristics supply the program with a “best guess” for what path the solution lies on. Page 10 Artificial Intelligence For many problems, it is possible to begin the search with some form of a guess and then refine the guess incrementally until no more refinements can be made. Other optimization algorithms are simulated annealing, beam search and random optimization in the field of study of the artificial intelligence. Evolutionary computation uses a form of optimization search. Form of evolutionary computation includes swarm intelligence algorithms and evolutionary algorithms. 5.3 Logic Logic was introduced into AI research by john McCarthy in his 1958 advice taker proposal. Logic is used for information representation and problem solving, but it can be applied to other problems as well in programming. Several different forms of logic are used in AI research. Propositional or sentential logic is the logic of statements which can be true or false. First order logic also allows the use of quantifiers and predicates, and can express facts about objects, their properties, and their relations with each other. Fuzzy logic is a version of first order logic which allows the truth of a statement to be presented as a value between 0 and 1, rather than simply true or false. Fuzzy systems can be used for uncertain reasoning and have been widely used in modern industrial and consumer product control systems. Subjective logic models uncertainty in a different and more explicit manner than fuzzy-logic. A given binomial opinion satisfies within a beta distribution. By this method, ignorance can be distinguished from probabilistic statements that an agent makes with high confidence. evasion logics, non monotonic logics and circumscription are forms of logic designed to help with evasion reasoning and the qualification problem. Several extensions of logic have been Page 11 Artificial Intelligence design to handle specific domains of knowledge, such as explanation logics, situation calculus and fluent calculus, casual calculus, and modal logic. 5.4 Classifiers and statistical learning methods A classifier can be trained in various ways there are many statistical and machine learning approaches. The simplest AI applications can be divided into two types’ classifiers and controllers. Controllers do however also classify conditions before inferring actions, and therefore classification forms a central part of many AI systems. Classifiers are functions that use pattern matching to determine a closest match. They can be tuned according to examples, making them very attractive for use in AI. These examples are known as observations or patterns. In supervised learning, each pattern belongs to a certain received, that observation is classified. The most widely used clarifies are the neural network, kernel methods such as the support vector machine, k-nearest neighbor algorithm, Gaussian mixture model, immature Bayes classifier, and decision tree. Classifier performance depends greatly on the characteristics of the data to be classified. There is no single classifier that works best on all given problems this is also referred to as the no free lunch theorem. Determining a suitable classifier for problem is still more an art than science. The performance of these classifiers has been compared over a wide range of tasks. 5.5 Evaluating progress A quite different approach measures machine intelligence through tests which are developed from mathematical definitions of intelligence. The main categories of networks are acyclic or feed forward neural networks and recurrent neural networks. Among the most popular feed forward networks are perceptions, multi-layer perceptions and radial basis networks. Page 12 Artificial Intelligence Neutral networks can be applied to the problem of intelligent control or learning, using such techniques as Hebbian learning and competitive learning. The broad classes of outcome for an AI test are: Sub-human: Perform worse than most humans Strong super-human: Performs better than all humans Super-human: performs better than most humans Optimal: it is not possible to perform better This procedure allows almost all the major problems of artificial intelligence to be tested. However, it is a very difficult challenge and at presents all agents fail. 6.0 Applications Artificial intelligence has successfully been used in a wide range of fields including medical diagnosis, stock trading, robot control, law, scientific discovery, video games, toys, and web search engines. Frequently when a technique reaches mainstream use, it is no longer considered artificial intelligence, sometimes described as the AI effect. It may also become integrated into artificial life. 6.1 Game playing You can buy machines that can play master level chess for a few hundred dollars. There is some AI in them, but they play well against people mainly through monster force computation looking at hundreds of thousands of positions. To beat a world champion by monster force and known unswerving heuristics requires being able to look at 200 million positions per second. Page 13 Artificial Intelligence 6.2 Speech recognition In the 1990s, computer speech recognition reached a practical level for limited purposes. Thus a united airline has replaced its keyboard tree for flight numbers and city names. It is convenient. On the other hand, while it is possible to instruct some computers using speech, most users have gone back to the keyboard and the mouse as still more convenient. 6.3 Understanding natural language Just getting a sequence of words into a computer is not enough. Parsing sentences is not enough either. The computer has to be provided with an understanding of the domain the text is about, and this is presently possible only for very limited domains. 6.4 Computer vision The world is composed of three-dimensional objects, but the inputs to the main human eye and computers cameras are two dimensional. Some useful programs can work solely in two dimensions, but full computer vision requires partial three-dimensional information that is not just a set of two- dimensional views. At present there are only limited ways of representing three-dimensional information directly, and they are not as good as what humans evidently use. 6.5 Expert systems A knowledge engineer interviews experts in a certain domain and tries to embody their knowledge in a computer program for carrying out some task. How well this works depends on whether the intellectual mechanisms required for the task are within the present state of AI. When this turned out not to be so, there were many disappointing results. One of the first expert systems was MYCIN in 1974, which diagnosed bacterial infections of the blood and suggested Page 14 Artificial Intelligence treatments. It did better than medical students or practicing doctors, provided its limitations were observed. Namely, its ontology included bacteria, symptoms, and treatments and did not include patients, doctors, death, recovery, etc., it is clear that the knowledge engineers forced what the experts told them in to a predetermined frame work. In the present state of AI, this has to be true. The usefulness of current expert systems depends on their users having common sense. 6.6 Heuristic classification One of the most feasible kinds of expert system given the present knowledge of AI is to put some information in one of a fixed set of categories using several sources of information. An example is advising weather to accept a proposed credit card purchase. Information is available about the owner of the credit card, his record of payment and also about the item he is buying and about the establishment from which he is buying it. 7.0 Uses: Some computer programs that are used to perform AI tasks are designed to manipulate symbolic information at extremely high speeds, in order to compensate for their partial lack of human knowledge and selectivity. Such programs are usually called “expert systems”. The software systems of this type that have been produced so far are limited in their vocabulary and knowledge to specific, narrowly defined subject areas. Knowledge-based expert systems enable computers to make decisions for solving complicated nonnumeric problems. These expert systems consist of hundreds or thousands of “if-then” logic rules formulated with knowledge gleaned from leading authorities in a given field. Programs have also been developed that enable computers to comprehend commands in a natural language ordinary English. They contain large amounts of information about the meaning Page 15 Artificial Intelligence of words pertaining to that subject, as well as information about grammatical rules and common violations of those rules. Other programs are designed to simulate human capabilities for problem solving through the use of highly selective search and recognition methods, rather than through superhuman processing speeds. Both experts systems and programs simulating human methods have attained the performance levels of human experts and professionals in performing certain specific tasks, but by the mid 1990s there were still no programs that could match human flexibility over wider domains or in tasks requiring much everyday knowledge. The ability to identify graphic patterns or images is associated with artificial intelligence, since it involves both cognition and abstraction. In a system designed with this capability, a device linked to a computer scans, senses, and transforms images in to digital patterns, which in turn are compared with patterns stored in the computer’s memory. The stored patterns can represent geometric shapes and forms that the computer has been programmed to identify. The computer processes the incoming patterns in rapid succession, isolating relevant features, filtering out unwanted signals, and adding to its memory any new patterns that deviate beyond a specified threshold from the old and are thus perceived as new entities in the area. 8.0 Future: Major and continuing advances in computer processing speeds and memory sizes have facilitated the development of AI programs. Although most AI programs attempting to simulate Page 16 Artificial Intelligence higher mental functions incorporate the restricted access of limited short term memory, which restricts humans to carrying out one or a few mental tasks at a time, many investigators have begun to explore how the intelligence of computer programs can be enhanced by incorporating parallel processing, the instantaneous execution of several separate operations by means of computer memories that allow many process to be carried out at once. The question of which portions of the human brain operate serially and which operate in parallel has been a topic of intense debate by researchers in both the cognitive sciences and AI, but no clear verdict had been reached by the mid 1990s. AI research has thus focused on understanding the mechanisms involved in human mental tasks and on deceitful software that performs similarly, starting with relatively simple ones and continually progressing to levels of greater involvedness. Even as AI technology becomes integrated into the fabric of everyday life, AI researchers remain focused on the grand challenges of automating intelligence Work is progressing on developing systems that converse in natural language, that perceive and respond to their surroundings, and that encode and provide useful access to all of human knowledge and expertise. Understanding of human intelligence, abstract understanding of intelligence continues to have practical consequences in the form of new industries, enhanced functionality for existing systems, increased productivity in general, and improvements in the quality of life. But the ultimate promises of AI are still decades away, and the necessary advances in knowledge and technology will require a sustained fundamental research effort. Page 17 Artificial Intelligence REFERENCES: Bruce G, Buchanan. (Winter 2005) “ A (Very) Brief History Artificial intelligence” Shapiro, Stuart C. (1992) “Artificial intelligence” Serenko, Alexander; Deltor, Brain (2004) “Intelligent agents as innovations” AI and society. Bernard, Ettienne (2006) “Data characteristics that determine classifier performance” Norvig, peter (2003), Artificial Intelligence: A modern approach, Upper saddle River, New Jersey: Prentice Hall, ISBN 0-13-790395-2 Page 18