Yapay zeka ve uygulamaları

Özkul, A. Semih
Süreli Yayın başlığı
Süreli Yayın ISSN
Cilt Başlığı
Fen Bilimleri Enstitüsü
İnsan zekasının anlaşılmasıyla zeki ve -Faydalı makineler yapmayı konu edinmiş Yapay Zeka çalışmalarının son ürünleri kendini her alanda hissettirmeye başlamış ve Yapay Zeka uygulamaları ard arda gelmeye başlamıştır. Yapay Zekanın son derece büyük olan potansiyel -Faydası.göz önüne alındısında, Yapay Zeka araştırma çalışmalarının önemi ortaya çıkmaktadır. Bu tez çalışmasında Yapay Zekanın tanımı, metot ve teknikleri, ilgi alanları ve -faydaları incelenmiştir. Bununla beraber bu çalışmanın bazı temel özellikleri şöyle sıralanabilir: i) Programlama tekniklerine büyük ağırlık verilmiştir, ii) Uygulamaya dönük bir yaklaşım takip edilmiştir. Teori ile birlikte pek çok örnek verilmiş, ayrıca hemen her bölümde bir ya da birden çok uygulama yapılmıştır. Bu açıdan çalışmada birden çok uygulama mevcuttur. Bölüm 1 ' de YZ'nın ne olduğu, tarihçesi, tanımı ve alanları belirtilmiştir. Bölüm 2'de Prolog dilinin temel özellikleri örneklerle anlatılmıştır. Bölüm 3' de veri yapıları incelenmiş, listeler, ağaçlar, graflar ve arama metotları anlatılmıştır. Bölüm 4' de höristik kavramı ve YZ için önemi ele alınmıştır. Bölüm 5' de çıkarım, mantık ve belirsizlik konuları işlenmiştir. Bölüm 6' da konuşma algılaması tanımı, aşamaları ve günümüzdeki kullanım alanları incelenmiştir. Bölüm 7' de konuşmanın yeniden üretilmesi ele almış ve uygulamalar yapılmıştır. Bölüm 8' de doğal insan dili incelenmiş ve Türkçe bir doğal dil işlemcisi geliştirilmiştir. Bölüm 9' da endüstriyel robotlar ele alınmış ve bir robot simülasyonu yapılmıştır. Bölüm 10' da makine görüntü algılama konusu incelenmiş ve bir uygulama yapılmıştır. Bölüm 1 de uzman sistemler tanımı, kullanım alanları, ve faydaları ile ele alınmış ve bir de uygulama yapılmıştır. Bölüm 12' de ise Yöneylem Araştırması ve Yapay Zeka arasındaki ilişkiler ve ortak noktaları ele alınmış, birbirlerine yapabilecekleri katkılar sıralanmış ve bir dinamik programlama uygulaması verilmiştir.
Artificial Intelliaence And Its Applications Artificial Intelligence (AI), is an emerging technology that has recently attracted considerable publicity. Many applications are now under development. One simple view of AI is that it is concerned with divising computers programs to make computers smarter. Thus research in AI is -Focused on devoloping computational approaches to intelligent behaviour. This research has two goals, i) making machines and computational process more useful, ii) understanding intelligence. AI is concerned in intelligent behaviour, primarily with non-numeric processes that involve complexity, uncertainty, and ambiguity and -For which known algorithmic solutions do not usually exist. Unlike conventional programming, it is knowledge based, almost in variably involves search. and uses heuristics to guide the solution process. Thus AI can be considered to be built on the -Following: 1) Knowledge o-F the domain o-F interest. 2) Methods -For operating on the knowledge. 3) Controls structures for choosing the appropriate method and modifying the database as required. AI, has both borrowed from and contributed to other closely related disciplines concerned with advanced methods for information processing. Thus, link exist between AI and aspects of such theoritical areas as mathematical logic, operations research, decision theory, information theory, pattern recognition and mathematical linguistics. While the general objective of AI work is the development of machines which exhibit intelligent behaviour, this is in itself too broad and ambigious to be meaningful. The following paragraphs will describe some of the areas of AI which are presently being pursued as distinct areas of research. XI A) Problem salving. Examples o-f problem salving systems are the chess plavina programs. Given a set of rules and strategies these systems are capable of playing chess on a proficient level. Problem solvina does not necessarily restrict itsel-F to games. Consider, in the -Future, telling your household robot to -Fetch the paper. On the wav to salving the problem of retreiving the paper, the robot must identi-Fy and solve a number o-f smaller problems. hopeful ly in the most efficient sequence. The techniques of problem solving are the same whether developing- a robotic assembly strategy solver or some other problem salver. B) Natural language. In spite of the proliferation of computers into department stores, many people are still uneasy and uncanfartable dealing with computers. This due in part to the need to talk for a computer in its language rather than the user's natural language. The problems facing natural researchers are numerous. The computer must not only be able to understand the meaning of words. but how those meaning may differ in context with other words. The systems must also be able to understand the syntax of the language so that the relationship of the words is understood. C) Expert systems. This area of research is concerned with developing systems that appear to behave as human experts in specific fields. Through a dialog with a human operator, an expert system can recommend test to be performed and ask appropriate auestions until it arrives at a conclusion. At present, expert systems are -being used to configure computer systems, design circuits, and perform medical diagnosis. Some of the issues involved with the design of expert systems include the problems of deaiina with vast amounts of data, explaining the system's reasoning on the way to reaching a conclusion, representing the data collected from the human experts, and improving the knowledge base with experience. D) Learning. One of the attributes of intelligence is the ability to learn from experience. If machines were capable of learning then the task of endowing them with knowledge, as in the case of expert systems, would be greatly simplified. Same systems has been developped which have shown the ability to learn from experience, but to this date limited progress has been made. E) Vision. Most of the basic concepts imoloved in commercial vision systems are the result of AI research. One of the more interesting goals of AI vision research is to permit the systems to perform scene analyses. That is, present the vision system with a scene and allow the system to identify objects within the scene. XXI Same of the other areas of AI research include: F) Robotics is an aoolied enaineerina science that has been referred to as a combination o-f machine tool tecnnoiogy and computer science. It includes such seemingly diverse fields as machine design, control theory, microelectronics, computer programming, human factors and production theory. Research and development are proceeding in all of these areas to improve the wav robot work and think. It is likelv that the research e-F-forts will result in -future robots that will make today's machines seem auite primitive. Advancements in technology will enlarge the scope of the industrial applications of robot. G) Speech recognition is our fastest means of discourse communication, being about twice as fast as the average typist. It is also nearly effortless: Speech does not need visual or physical contact and it places few restricitons on the use of hands or the mobility of the body. Speech is thus well suited to communication with a machine when a person is engaged in other activities. As a result, speech recognition and understanding is one of the most difficult AI subject, being a perception task related to the scene understanding problem in computer vision. Although the constraining aspects of natural language help reduce the magnitude of the task, it remains a major problem area. H) Speech synthesis (production) is. an emerging technology whose products are already becoming common place. Although the present market for these devices is still small. The future looks very bright. Speech production is not narmally considered an AI topic. Although it is sure to play an important part in many future AI systems, particularly when coupled with speech understanding. One may very well consider these production systems, which employ rules (often heuristics) for driving speech from stored speech elements, as an exemple of an expert system on a chip. It appears that technology is now ready for early applications. However, the fact that current Artificial Intelligent systems are prone to error suggest that current Artificial Intelligence applications should be focused on intelligent aids for human beings rather than on truly autonomous systems. The following list is about overall conclusions on the current state of the art in AI: - Basic principles and techniaues devised and demonstrated. - Inital language, programs and tools devolopped. - Software portability is a problem. xiii - Same initial applications already in use. - Technoloav is now readv -Far earlv appl ications. - Current technology more appropriate -For intelligent assistants than for autonomous svstems. - Customizing adapting and usually writing own programs necessary. - Because of huge potentials benefits, utilization will be exolosive as technoloav is -Further rationalized. There is a strong need -For supporting more teaching and research in AI at universities. This can be aided bv government programs and grants -from large corporations. It can also be -facilitated by encouraging AI personnel in industry to teach part time. In this master tnesis. the definition. fields of interest and techniques of Artificial Intelligence have been introduced. Especially programming techniques have been widely studied, and an application based approach has been fol lowed. In Chapter 1, the definition, history and fields of interest has been introduced. In Chapter 2, the fundamentals of prolog programming languages has been explained. In Chapter 3, data representation structures such as lists, trees, graphs and search methods has been studied. In Chapter 4, the heuristics concept has been explained, and its importance for AI has been streessed. In Chapter 5, reasoning, logic, predicate logic and uncertainty has been studied. In Chapter 6, the definition, levels, and today's usage of speech recognition has been presented. In Chapter 7, speech production has been studied, its levels, benefits and fields has been given. In Chapter 8, natural language parsers has been explained, and some English language parsers and a Turkish parser have been developed. In Chapter 9, robotics field has been introduced. Especially, industrial robots have been studied and a robot simulation program has been developed. XIV In Chapter 10, machine vision has been studied with oattern recognition. In Chapter 11, expert systems have been introduced. Its de-Finition, -Fields and benefits have been given, and an application has been developed. In Chapter 12. the relationship between Artificial intelligence and Operations Research has been discussed, and a dynamic programming application written in Prolog has been given.
Tez (Yüksek Lisans) -- İstanbul Teknik Üniversitesi, Fen Bilimleri Enstitüsü, 1991
Anahtar kelimeler
Endüstriyel robotlar, Robotlar, Sezgisel, Uzman sistemler, Yapay zeka, Industrial robots, Robots, Intuitive, Expert systems, Artificial intelligence