Artificial Intelligence - Report Example

Artificial Intelligence Part I For each of the following questions, determine if the answer is true (T) or false (F Logic, rules, semantic networks and frames are major alternatives for knowledge representation. (T) 2. The major concern of logic is the development of formal representation languages with sound and complete inference rules. (T) 3. Common sense reasoning and formal logic can be made to directly correspond to each other. (T) 4. Logical implications only express a relationship between the truth values of their operands. (T) 5. Associationist theories define the meaning of an object in terms of a network of associations with other objects. (T) 6. Expert systems have been a greater driving force for research in knowledge representation than natural language understanding. (T) 7. A script is a structured representation describing a usual sequence of events in a particular context. (T) 8. Scripts can be used as a means of organizing conceptual dependency structures into descriptions of typical situations. (F) 9. Frames is an example of a network representation language. (T) 10. Conceptual graphs is an example of a network representation language. (F) 11. The phenomenon of knowledge in the agent-oriented environment is non-emergent, i.e. knowledge is just a property of an individual agent. (T) 12. According to Luger, the four criteria for an intelligent agent system include software problem solvers that are situated, autonomous, inflexible and asocial. (T) 13. Planning has applications in manufacturing, robotics and expert systems. (T) 14. In expert systems solutions reasoning not just about the plan but also about the process of planning is important. (T) 15. Neurally inspired approaches to learning, e.g. neural networks are also know as serial distributed processing systems or disconnectionist systems. (F) 16. Neural networks de-emphasize the use of symbols in problem solving as they purport that intelligence arises in systems of simple, interacting components through a process of learning which adjusts the connections between components. (T) 17. The basis of neural networks, the artificial neuron, consists of input signals, a set of real-valued weights, and activation level and a threshold function. (T) 18. The network topology, the learning algorithm and a motivation function are considered characterizing global properties of neural networks. (T) 19. LISP and PROLOG are two of the main programming languages used in AI today. (T) 20. Refutation is the proof procedure that is the primary method for computing with PROLOG. (F) 21. PROLOG uses the propositional calculus and an interpreter executes the program by systematically making inferences from logic specifications. (T) 22. LISP is a programming language based on the theory of recursion. (T) 23. In LISP, the list is the basis of both programs and data structures. (T) 24. LISP is an imperative language, i.e. LISP programs describe how to perform an algorithm. (T) 25. PROLOG is a declarative language, i.e. PROLOG programs are assertions that define relationships and constraints in a problem domain. (T) 26. There are AI scientists today who argue that the traditional viewpoint of the individual mind as the sole source of intelligence is not sufficient; they contend that we need to consider knowledge as a social construct also. (T) 27. Good artificial intelligence models are never falsifiable. (F) 28. There are large philosophical implications of artificial intelligence in addition to the engineering techniques used to build intelligent computer-based objects. (T) 1. What are three of the major accomplishments in the field of artificial intelligence to-date? Discuss your answer and include details and explanation. The three major accomplishments in the field of artificial intelligence are Neural Nets and Genetic Algorithms, Natural Language Understanding and Semantics, and Human Performance Modeling. Neural Nets and Genetic Algorithms represent the approach to build intelligent programs on the basis of models that are parallel to the structure of neurons in the human brain or the emerging patterns that are seen in genetic algorithms and artificial life. In neural architecture each computational unit computes some function of the inputs provided and passes the result to the other connected units of the network, and thus the end result is got by parallel and distributed processing within the network of neural connections and threshold weights. Genetic algorithms and artificial life enable the finding of new solutions by making use of components of previous solutions. The genetic operators like crossover and mutation work to generate for every new generation even better solutions for potential problems. Artificial strives to create a new generation as a function of quality with regard to its neighbors of previous generations. A fundamental aspect of human intelligence is the ability to use and understand natural language. Artificial intelligence attempts to develop the same ability through Natural Language Understanding and Semantics. Successful automation of the ability to use and understand natural language would have a profound effect on the usability and effectiveness of computers. Human Performance Modeling gives artificial intelligence a very large part of its basic methodology. In addition it is also a powerful tool for formulating and testing theories of human cognition, so much so several psychologists have adopted the language and theory of computer science to create models of human intelligence. This has enabled psychologists to empirically test, critically evaluate, and refine their ideas. (1) 2. Based on your readings and research this semester discuss where you think the future of artificial intelligence lies? Which areas of artificial intelligence will make the most progress? Why? What do you feel will be accomplished in our lifetime. Include factual information in your response. In the immediate future developments in the field of artificial intelligence would see greater emphasis and development of artificial intelligence convenience products for humans. These products would mean less of the requirement for humans to be involved in domestic chores, navigational and driving aids that require lesser human intervention in the driving of a car Moving from the physical labor part artificial intelligence in the future would provide the means of studying of stock markets, managing domestic accounting, and searching the Internet for sections of interest New drugs would emerge in the market as a result of artificial intelligence. These are just some of the possibilities. The areas of artificial intelligence where maximum progress could be expected are in the field of robotics and the search for the means to mimic the human brain. Robotics will continue to flourish for two reasons. The first is the utility factor in many areas of human activity. The second is robots have caught the fancy of humans. Both these factors contribute to increasing business enterprise interest in the field of robotics. The ultimate goal of artificial intelligence would be to match human intelligence. The computing power of computers is growing at quite a rapid pace, and so is the penetration of computers into human activity around the world. This should assist in further enhancements in the abilities of artificial intelligence. However during our lifetime there is no way that artificial intelligence will catch up with human intelligence, as just in the area of computing power the human brain remains far ahead. In the field of robotics during our lifetime more advanced robotics capable of increased human-like functions can be expected. (2) 3. Discuss neural networks. Include in your discussion the properties of individual neurons as well as the global properties of neural networks. Also include in your discussion why artificial intelligence scientists are interested in neural networks and what their significance is to the field of AI. A computer network that has been modeled after the cortical structures of the human brain is termed a neural network, with each of the interconnected processing elements called neurons. In a neural network each neuron computes some function of the inputs provided and passes the result to the other connected units of the network, and thus the end result is got by parallel and distributed processing within the network of neural connections and threshold weights. The simple processing neurons demonstrate global behavior on the basis of two factors namely the connections between the processing elements and the element parameters. The significance of neural networks to the field of artificial intelligence lies in their advantages over traditional artificial programs. Traditional artificial intelligence programs tend to be brittle, and are hypersensitive to. Human intelligence being more flexible is good at interpreting noise. An example of this is the ability of human intelligence to discern a face in a darkened room or a particular conversation at a noisy party. Neural networks that mimic the human brain display a high potential for matching noisy and incomplete data. He reason for this feature in neural networks is because they capture knowledge in a great number of individual units distributed in the network. (1) 4. Compare LISP and PROLOG. Include in your discussion the major features of each language and why both are considered important in artificial intelligence. If you were going to design an AI programming language what features would you include in the language and why would you want them? LISP stands for list programming and is the second oldest computer programming language. LISP offers a means for functional programming in artificial intelligence. PROLOG is a subsequently developed programming language and provides a means for logic programming. The core to every LISP programming lies in the interpreter, which computes a value for a given s-expression. This process is known as evaluation. LISP in reality possesses operational semantics, but with a precise mathematical definition derived from recursion function theory. Programming in Lisp really is defining functions that operate on lists. As this is the sore of LISP, every LIST system comes with a basic set of primitive built-in functions that effectively support the main list operations. Prolog consists of a method for specifying predicate calculus propositions, and their relationships, and rules specifying their logical relationships. Programming in Prolog is based on the specification of facts about objects, and their relationship, and rules specifying their logical relationships. Prolog programs are declarative collections of a problem, as they do not specify how a result is to be computed, but rather define what the logical structure should be. It is in this that the PROLOG is different to LISP, where the focus is on how a result is top be computed. Using PROLOG, programming can be done at a very abstract level quite close to the formal specification of a problem. If I were to choose a programming language I would choose a logic program language, as it would give me the flexibility of defining the logical structure of the solution to a problem, and expect a result for it. This would enable me to do programming at an abstract level near to the formal specification of the problem. (1) Works Cited 1. Luger, George & Stubblefield, A. William. (1993). Artificial intelligence : structures and strategies for complex problem-solving. Second Edition. Benjamin-Cummings. California. 2. Moy, Chris. “The Future of Artificial Intelligence”. sffworld.com. 2000. 11 Dec. 2006. . Read More

Discuss neural networks. Include in your discussion the properties of individual neurons as well as the global properties of neural networks. Also include in your discussion why artificial intelligence scientists are interested in neural networks and what their significance is to the field of AI. A computer network that has been modeled after the cortical structures of the human brain is termed a neural network, with each of the interconnected processing elements called neurons. In a neural network each neuron computes some function of the inputs provided and passes the result to the other connected units of the network, and thus the end result is got by parallel and distributed processing within the network of neural connections and threshold weights.

The simple processing neurons demonstrate global behavior on the basis of two factors namely the connections between the processing elements and the element parameters. The significance of neural networks to the field of artificial intelligence lies in their advantages over traditional artificial programs. Traditional artificial intelligence programs tend to be brittle, and are hypersensitive to. Human intelligence being more flexible is good at interpreting noise. An example of this is the ability of human intelligence to discern a face in a darkened room or a particular conversation at a noisy party.

Neural networks that mimic the human brain display a high potential for matching noisy and incomplete data. He reason for this feature in neural networks is because they capture knowledge in a great number of individual units distributed in the network. (1) 4. Compare LISP and PROLOG. Include in your discussion the major features of each language and why both are considered important in artificial intelligence. If you were going to design an AI programming language what features would you include in the language and why would you want them?

LISP stands for list programming and is the second oldest computer programming language. LISP offers a means for functional programming in artificial intelligence. PROLOG is a subsequently developed programming language and provides a means for logic programming. The core to every LISP programming lies in the interpreter, which computes a value for a given s-expression. This process is known as evaluation. LISP in reality possesses operational semantics, but with a precise mathematical definition derived from recursion function theory.

Programming in Lisp really is defining functions that operate on lists. As this is the sore of LISP, every LIST system comes with a basic set of primitive built-in functions that effectively support the main list operations. Prolog consists of a method for specifying predicate calculus propositions, and their relationships, and rules specifying their logical relationships. Programming in Prolog is based on the specification of facts about objects, and their relationship, and rules specifying their logical relationships.

Prolog programs are declarative collections of a problem, as they do not specify how a result is to be computed, but rather define what the logical structure should be. It is in this that the PROLOG is different to LISP, where the focus is on how a result is top be computed. Using PROLOG, programming can be done at a very abstract level quite close to the formal specification of a problem. If I were to choose a programming language I would choose a logic program language, as it would give me the flexibility of defining the logical structure of the solution to a problem, and expect a result for it.

This would enable me to do programming at an abstract level near to the formal specification of the problem. (1) Works Cited 1. Luger, George & Stubblefield, A. William. (1993). Artificial intelligence : structures and strategies for complex problem-solving. Second Edition. Benjamin-Cummings. California. 2. Moy, Chris. “The Future of Artificial Intelligence”. sffworld.com. 2000. 11 Dec. 2006.

Read More