yvsou.com

Artificial Intelligence: Application and Future Prospect

PON PISETH 2016529620001

（浙江理工大学信息学院，杭州310000）


 Abstract: The world in twentieth first century is absolutely different from what it is used to be in the twentieth century and of all times of the world history. In the last few decades the world has changed dramatically because of engineering developments technological innovation and scientific discoveries. On the vision of technology world artificial intelligence is improving and spreading the wide range of its potential to be the world’s leading technology. Recently most of developed countries has spent enormous amount of budgets for the research in the field of artificial intelligence to seek even more effective technology which will solve the problems we have risked in many sectors of our daily life. Although methods being used in AI are likely to be full of abstract and complexity but engineers are still optimistic about the future of AI that will influence our world in the very near future. This paper will give short introduction of artificial intelligence describe about the relationship between part of artificial intelligence and other fields its tremendous application and its important role as well as its future prospect that will be discussed in the end of the paper.
 Keywords: cognitive manipulation neural neuroscience reasoning semantic syntactic.

<h1 style="text-align: left;">I. Introduction
Artificial Intelligence (AI) is the field of computer science the sub-discipline of computing that seeks to build autonomous machines- machines that can carry out complex tasks without human intervention. To many people it represents the future of computing but to others it is an avenue for applying new and different technologies to problem solving.
The term artificial intelligence probably conjures up various images in your mind such as a computer playing chess or a robot doing household chores. These are certainly aspects of AI but it goes far beyond that. AI techniques affect the way we develop many types of application programs from the mundane to the fantastic. The world of artificial intelligence opens doors that no other aspect of computing does. Its role in the development of state-of-the-art application programs is crucial. AI successfully understands human speech compete at a high level in strategic game systems self-driving cars intelligent routing in content delivery networks and interpreting complex data.
AI research is divided into subfields that focus on specific problems or on specific approaches or on the use of a particular tool or towards satisfying particular applications. The central problems of AI research include search method computing decision making reasoning knowledge representation planning learning natural language processing and pattern recognition etc.
In the twentieth first century AI techniques both hard and soft have experienced a resurgence following concurrent advances in computer power sizes of training sets and theoretical understanding and AI techniques have become an essential part of the technology industry helping to solve many challenging problems in computer science.

<h1 style="text-align: left;">II. Part of Artificial Intelligence
<h2 style="text-align: left;">1. Search Methods and Evolutionary Computation
Many problems in AI can be solved in theory by intelligently searching through many possible solutions. Search methods are among the earliest methods of AI. It can be used versatility for chronically reasoning and for constructing AI systems. A search tree is a structure that represents all possible decisions that would be made. Its concept can apply nicely in such complicated games such as chess. Blind search is a concept that a solution would be searched by the systematic generation of new states and checking whether the solution is founded by chance. Heuristic search allows us to expand the paths that seem to be the most promising. Its method can be used if we are able to define a heuristic function that estimates how far a search tree node is from a solution.
Evolutionary Computation uses a form of optimization search. Forms of evolutionary computation include swarm intelligence algorithms (such as ant colony or particle swarm optimization) and evolutionary algorithms (such as genetic algorithms gene expression programming and genetic programming).

<h2 style="text-align: left;">2. Logic-based Reasoning
Logic is used for knowledge representation and problem solving and 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 represented as a value between 0 and 1 rather than simply TRUE (1) or FALSE (0). 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 belief + disbelief + uncertainty=1 within a Beta distribution. By this method ignorance can be distinguished from probabilistic statements that an agent makes with high confidence.
Default Logics non-monotonic logics and circumscription are forms of logic designed to help with default reasoning and the qualification problem. Several extensions of logic have been designed to handle specific domains of knowledge such as: description logics; situation calculus event calculus and fluent calculus (for representing events and time); causal calculus; belief calculus; and modal logics.


<h2 style="text-align: left;">3. Structural Models of Knowledge Representation
Structural models of knowledge representation are used for defining declaration knowledge. They are usually in the form of graph-like hierarchical structures. The knowledge we need to represent an object or event varies based on the situation. Depending on the problem we are trying to solve we need specific information. We also need particular information in a form that allows us to search and process that information efficiently.
In general we want to create a logical view of the data independent of its actual underlying implementation so that it can be processed in specific ways. In the world of AI the information we want to capture often leads to new and interesting data representations. We want to capture not only facts but also relationships. The kind of problem we are trying to solve may dictate the structure we impose on the data. As specific problem areas have been investigated new techniques for representing knowledge have been developed.
Semantic network is a knowledge representation technique that focuses on the relationships between objects. A directed graph is used to represent a semantic network or net. The nodes of the graph represent objects and the arrows between nodes represent relationships. The arrows are labeled to indicate the types of relationships that exist. There are essentially no restrictions on the types of relationships that can be modeled in a semantic network. A semantic network is a powerful versatile way to represent a lot of information. The challenge is to model the right relationships and to populate the network with accurate and complete data.
frames can be treated as a substantial extension of semantic networks. A node of such a network is called a frame and has a complex internal structure. It allows one to characterize objects and classes in a detailed way. frame theory has the following psychological observation as a basic assumption. If somebody encounters a new unknown situation then he/she tries to get a structure called a frame out of his/her memory. This structure which represents a stereotyped situation similar to the current situation can be then used for generating an adequate behavior. In AI systems a graph- like structure of frames is defined according to precise principles which allow the system to process and analyze it in an automatic way.
script is a method for Natural Language Processing (NLP) which can be defined with the help of conceptual dependency graphs. The model is also based on a certain observation in psychology. If one wants to understand a message concerning a certain event then one can refer to a generalized pattern related to the type of this event. The pattern is constructed on the basis of similar events that one has met previously. Then it is stored in the human memory. One can easily notice that the concept of scripts is similar conceptually to the frame model.

<h2 style="text-align: left;">4. Syntactic Pattern Analysis
In syntactic pattern analysis reasoning is performed on the basis of structural representations which describe things and phenomena belonging to the world. A set of such structural representations called patterns constitutes the database of an AI system. This set is not represented in the database explicitly but with the help of a formal system which generates all its patterns. A generative grammar is the most popular formal system used for this purpose. The grammar generates structural patterns by the application of string rewriting rules which are called productions. Thus a string generative grammar constitutes a specific type of Abstract Rewriting System (ARS) which is called a String Rewriting System (SRS). Therefore reasoning by syntactic pattern analysis can be treated as reasoning by symbolic computation.

<h2 style="text-align: left;">5. Rule-based Systems
The main idea of reasoning in rule-based systems is in fact the same as in the case of logic-based reasoning. Both models are based on deductive reasoning. As a matter of fact the form of expressions which are used for knowledge representation is the main difference between these models. In the case of logic-based reasoning expressions are formalized considerably whereas in rule-based systems the expressions in the form of the so-called rules are represented in the following intuitive way: “If a certain condition is fulfilled then perform a certain action”. Additionally the way of formulating both a condition and an action is much easier to comprehend than in the case of FOL terms or expressions used in symbolic computing. This is of great importance for designing knowledge bases which are usually developed not only by IT specialists but also by experts in the field. Therefore clarity of expressions in knowledge base is recommended.

<h2 style="text-align: left;">6. Pattern Recognition and Cluster Analysis
In pattern recognition and cluster analysis various objects phenomena processes structures etc. can be considered as patterns. The notion is not limited to images which can be perceived by our sight. There are three basic approaches in the area of pattern recognition. In the approach based on a feature space a pattern is represented by a feature vector. If patterns are of a structural nature then syntactic pattern recognition or the structural approach is used.
In general pattern recognition consists of classifying an unknown pattern into one of several predefined categories called classes. Cluster analysis can be considered a complementary problem to pattern recognition. Grouping a set of patterns into classes categories is its main task.

<h2 style="text-align: left;">7. Neural Networks
Some artificial intelligence researchers focus on how the human brain actually works and try to construct computing devices that work in similar ways. An artificial neural network in a computer attempts to mimic the action of the neural networks of the human body. Each processing element in an artificial neural network in analogous to a biological neuron. An element accepts a certain number of input values and produces a single output value of either 0 or 1. These input values come from the output of other elements in the network so each input value is either 0 or 1. Associated with each input value is a numeric weight. The effective weight of the element is defined as the sum of the weights multiplied by their respective input values. The pathways established in an artificial neural net are a function of its individual processing elements. And the output of each processing element changes on the basis of the input signals the weights and the threshold values. But the input signals are really just output signals from other elements. Therefore we affect the processing of a neural net by changing the weights and threshold value in individual processing elements. The process of adjusting the weights and threshold values in a neural net is called training. A neural net can be trained to produce whatever results are required. Initially a neural net may be set up with random weights threshold values and initial inputs. The results are compared to the desired results and changes are made. This process continues until the desired results are achieved.
Neural Networks have been used successfully in thousands of application areas in both business and scientific endeavors. They can be used to determine whether an applicant should be given a mortgage. They can be used in optical character recognition allowing a computer to read a printed document. They can even be used to detect plastic explosives in luggage at airports.

<h2 style="text-align: left;">8. Reasoning with Imperfect Knowledge
If we reason about propositions in AI systems which are based on classic logic we use only two possible logic value true and false. However in the case of reasoning about the real physical world such a two-valued evaluation is inadequate because of the aspect of uncertainty. There are two sources of this problem: imperfection of knowledge about the real world which is gained by the system and vagueness of notions used for describing objects/phenomena of the real world.
There are three aspects of the imperfection of knowledge: uncertainty of knowledge (information can be uncertain) imprecision of knowledge (measurements of signals received by the AI system can be imprecise) and incompleteness of knowledge (the system does not know all required facts).
The model of Bayesian inference based on a probability measure. This measure is used to express our uncertainty concerning knowledge not for assessing the degree of truthfulness of propositions. Demister-Shafer theory which allows us to express a lack of complete knowledge for example our ignorance with specific measures. Various models of non-monotonic reasoning can also be applied for solving the problem of incompleteness of knowledge. Three such models namely default logic auto epitomic logic and circumscription reasoning are also being discussed.

<h1 style="text-align: left;">III. Application Areas of AI Systems
<h2 style="text-align: left;">1. Perception and Pattern Recognition
Intelligent behavior depends on perception of the external world to some extent. Although a human being perceives with the help of five senses i.e. sight hearing taste smell and touch only the first two senses are simulated in most AI systems. From a technical point of view both sound and image are treated one or two dimensional signals.
In AI systems the task of perceiving sound or image is divided into two main phases. The first phase concerns of receiving a corresponding signal with the help of a sensory device (camera or microphone) its preprocessing and its coding in a certain format. The methods used in this phase belong to conventional areas of computer science such as signal processing theory and image processing theory. Both theories were developed remarkably in the second half of the twentieth century. They allow us to implement systems which surpass human beings in some aspects of sensory perception.
In the second phase of perception in which sensory information is ingested thoroughly AI systems can use methods belonging to three groups of models that have been introduced in the monograph namely pattern recognition neural networks and syntactic pattern recognition. Let us notice that also in these areas a lot of efficient techniques have been developed. Optical Character recognition systems vision systems of industrial robots optical quality control systems in industry analysis of satellite images military object identification systems and medical image systems are some example of practical application of such systems. Image understanding is especially useful in the area of advanced medical diagnostics.
However there are still challenges in this area. Automatic learning is a crucial functionality of pattern recognition systems. In the case of classical pattern recognition and neural networks models contain adaptive techniques of learning. On the other hand in syntactic pattern recognition the issue of a system self-learning is more difficult since it relates to the problem of formal grammar induction. In this area research is still in a preliminary phase.

<h2 style="text-align: left;">2. Knowledge Representation
The problem of adequate knowledge representation has been crucial since the very beginning of developments in the AI area. An intelligent system should be able to adapt to its environment. Thus it should be able to acquire knowledge which describes this environment then to store this knowledge in a form allowing a quick and adequate intelligent response to any stimulus generated by the environment. Patterns of such responses represented as procedural knowledge should be stored in the system as well.
According to the first criterion knowledge representation models can be divided into the following three groups:
· Models of symbolic knowledge representation formulated in an explicit way. This representation triggers some knowledge i.e. formal grammars representations based on mathematical logic models in reasoning systems and schemes in case-based reasoning systems.
· Models of symbolic numeric knowledge representation formulated in an explicit way. These models are used if the notions which are the basis for the representation model are fuzzy they are ambiguous or imprecise.
· Models of knowledge representation formulated in an implicit way. This form is applied if knowledge is represented in a numeric way. It is typical for pattern recognition methods and neural networks. Such representations are of the form of clusters consisting of vectors sets of parameters (in pattern recognition) and weight vectors (in neural networks).

Where the second criterion i.e. the way of acquiring knowledge is concerned representation models can be divided into the following two groups:
· Models in which knowledge can be acquired by the system automatically.
· Models in which knowledge representation is defined and entered into the system by a knowledge engineer. Most models of knowledge representation formulated in an explicit way belong to this group.
Summing up automatic acquisition of knowledge in models based on symbolic knowledge is the crucial issue in this area. An automatic conceptualization is the main problem here and it has not been solved in a satisfactory way till now.

<h2 style="text-align: left;">3. Problem Solving
We define the area of problem solving as research into constructing generic methods that can be used for solving general problems. General Problem Solver (GPS) is a good example of this area of AI. The dream of AI researchers to construct such a system has not come true yet. Therefore this problem has been divided into a variety of sub problems such as reasoning decision making etc.
Systems based on a search strategy solve problems in cooperation way with a human designer. There are two phases of problem solving with the help of a search strategy:
· A phase of constructing an abstract model of the problem which is the basis for defining states in the state space
· A phase of searching the state space.
The development of methods which allow an AI system to autonomously construct an abstract model of a problem on the basis of perception of the problem seems to be one of the biggest challenges in the area of simulating cognitive or mental abilities.

<h2 style="text-align: left;">4. Reasoning
Artificial Intelligence systems work perfectly where deductive reasoning is concerned. Deductive reasoning is a type of reasoning in which on the basis of a certain general rule and a premise. System based on mathematical logic are the best examples of such reasoning. Reasoning systems are applied in business medicine industry communications transport etc. AI in the area of deductive reasoning we are able to simulate human abilities better than in the case of the remaining cognitive/mental abilities. This results from the dynamic development of mathematical logic in the first half of the twentieth century. Its models have been used successfully for defining effective algorithms of reasoning.

<h2 style="text-align: left;">5. Decision Making
Supporting a process of decision making was one of the first applications of AI systems. The natural approach based on a simulation of succeeding steps of a decision process performed by human expert is used in expert rule-based systems. The possibility of equating the set of possible decisions with the set of classes determined by the vectors of a learning set is a condition of using such an approach.
At the end of the twentieth century effective methods of decision making were developed on the basis of advanced models of decision theory game theory and utility theory. Decision support systems are applied in many application areas. Typical application areas include e.g. economics management medicine national defense and industrial equipment control.

<h2 style="text-align: left;">6. Planning
Planning consists of defining a sequence of activities which should result in achieving a predefined target. Simulation of this mental ability seems to be very difficult. It contains a crucial element of predicting consequences of taking certain actions. This task is especially difficult if it is performed in a real-time mode and in a changing environment which is typical in practical applications. Then a system has to modify very quickly a plan which has been already generated in order to keep up with the changing environment.
Planning in the real world is sometimes connected with some circumstances facts or situations which limit the possibility of our activity in the sense of time space other conditions related to the physicality of the world preferences concerning the way of achieving a goal etc. Then a planning problem can be expressed as a Constraint Satisfaction Problem CSP.
In the areas of AI planning problems are very important because of their various practical applications. Therefore many advanced methods which are based of such models as temporal logic dynamic logic situation calculus and interval algebra have been defined in this area recently.

<h2 style="text-align: left;">7. Natural Language Processing (NLP)
The research area of Natural Language Processing NLP should be divided into two subareas. The first subarea includes problems which can be solved by an analysis of a language on the syntactic and lexical level. For example text proofreading extraction of information from a text automatic summarizing Optical Character Recognition (OCR) speech synthesis simple question answer dialogue systems etc. belongs to this group. The second subarea contains problems which can be solved analysis of a language on the semantic level. This division has been introduced because nowadays only problems belonging to the second group are challenging in AI. For example automatic translation from a natural language into another natural language speech text understanding systems of human computer verbal communication etc.
Passing a message in one specific sense reveals the intention of its sender. He/ She passes this sense by stressing the proper phrase. However the ability to understand the correct sense of the message on the basis of stress intonation etc. relates to social intelligence. Although in this case we mean elementary social intelligence it is very difficult to embed this kind of intelligence in an AI system. Nevertheless Natural Language Processing can be considered a well-developed area of AI.

<h2 style="text-align: left;">8. Learning
Learning models in Artificial Intelligence can be divided into two basic groups:
· Experience generalization models which is connected with the behavioral approach in psychology. Such learning is treated as gaining experience which is done in order to modify the system’s behavior. However such learning can also be applied to symbolic representation. The scheme grammar induction- automation synthesis is a good example here. Then the response function f takes the form of a formal automation.
· Models transforming a representation of a problem domain correspond instead to models of cognitive psychology. In this case an AI system should construct a world representation. Then the system should transform it on the basis of the new knowledge gained. Thus a learning process can be divided into two phrases ontology construction and ontology transformation. In order to construct an ontology the system should firstly define concepts on the basis of observations of the world. Then it should define structures which describe semantic relations among these notions. Unfortunately AI systems are not able to perform such a task nowadays. However AI systems are able to learn by transforming ontologies predefined by a human designer. In this case the system extracts knowledge from the ontology by transformation operations.
Learning models which transform a problem domain have been developed dynamically in AI since the 1980s. The most popular methods include Explanation-based Learning Relevance-based Learning and Inductive Logic Programming.

<h2 style="text-align: left;">9. Manipulation and Locomotion
Since we do not remember this stage of our life well we do not realize the difficulty of acquiring these abilities. The simulation of these abilities is one of the most difficult problems in Artificial Intelligence strictly speaking in robotics which is an interdisciplinary research area making use of models of automatic control mechatronics electronics cybernetics and computer science.
Firstly manipulation and locomotion abilities of robots depend strongly on functionalities of other systems such as perception and pattern recognition systems problem solving systems or planning systems. Secondly manipulation and locomotion abilities of robots also depend on the technological possibilities of execution devices such as effectors actuators etc. Mobile robots for military or search-and-rescue applications are often constructed on the basis of the locomotion abilities of insects snakes or four-limbed animals not to mention intelligent aerial mobile robots and underwater drones. Generally in the area of locomotion constructors of mobile robots and devices have achieved amazing achievements recently.
Manipulation abilities of robots surpass those of humans in certain applications especially if high precision manual dexterity or high resistance to tiredness are required. Manipulation microsurgical robots and robots aiding microbiology experiments are good examples here. Of course these robots are tele-manipulators (or remote tele-manipulators) which are controlled by operators. Summing up there have been remarkable results in the area of intelligent manipulators and one can expect further successes in this field.
In spite of the fact that there are some interesting and usually spectacular results in the area of humanoid or android robotics we still await robots which can stimulate violin virtuoso or a prima ballerina.

<h2 style="text-align: left;">10. Social Intelligence Emotional Intelligence and Creativity
At the end of the twentieth century research into simulating both social intelligence and emotional intelligence in AI systems began. This has concerned synthetic aspects of the problem e.g. recognizing human mood on the basis of speech intonation. Stimulating human abilities in the analytic aspect is of course more difficult. In order to analyze facial expression and features of speech advanced pattern recognition methods are applied. Rule-based systems are used for the purpose of integrating vision and sound. Surely research in this area is very important since its results together with achievements in robotics can be applied in medicine social security etc. Distinct emotional messages sent by humans via e.g. facial expressions are recognized quite well nowadays by AI systems. Stimulation of transformational creativity in artificial systems is a really challenging task in the AI area.
Creative AI systems are implemented for solving general problems generating music and visual art etc. Various methods such as state space search neural networks genetic algorithms semantics networks and reasoning by analogy are used for these purposes.


<h1 style="text-align: left;">IV. Future Prospect of AI
<h2 style="text-align: left;">1. Issues of Artificial Intelligence
Let us begin our considerations with an analysis of the term artificial intelligence. In fact it has two basic meanings. Firstly it means a common research field of computer science and robotics in which development of systems performing tasks which require intelligence when performed by humans is a research goal. Secondly it means a feature of artificial systems which allows them to perform task that require intelligence when made by humans. Thus in this meaning artificial intelligence is not a thing but a property of certain systems just as mobility is a property of mobile robots which allows them to move. It is the subject of research in a discipline called cognitive science rather than computer science or robotics. Cognitive science is a new interdisciplinary research field on mind and cognitive processes concerning not only humans but also artificial systems. Its research focuses on issues which belong to philosophy psychology linguistics neuroscience computer science logic etc.
On the basis of artificial intelligence can be divided into the following two groups:
· Strong Artificial Intelligence which claims that a properly programmed computer is equivalent to a human brain and its metal activity.
· Weak Artificial Intelligence in which a computer is treated as a device that can simulate the performance of a brain. In this approach a computer is also treated as a convenient tool for testing hypotheses concerning brain and mental processes.
The behavior of systems can be explained on three levels of abstraction. At the lowest level called the physical stance and concerning both the physics and chemistry domains we explain the behavior of a system in a causal way with the help of the principles of science. The intermediate level called the design stance includes biological systems and systems constructed in engineering. We describe their behavior in a functional way. Minds and software belong to the highest level called the intentional stance. Their behavior can be explained using concepts of intentionality beliefs etc. However the concepts such as intentionality belief and mind do not explain anything. So they should be removed from science and replace with terms of biology and neuroscience.
Researchers who develop AI systems also take part in the discussion about Strong AI. Similarly to the case of philosophers and cognitivists views on this matter are divided. For some of them successes in constructing AI systems show that in the future the design of an Artificial Brain will be possible.

<h2 style="text-align: left;">2. Potential Barriers and Challenges in AI
A psychological definition of intelligence is a set of abilities which allow one firstly to adapt to a changing environment secondly a cognitive activity consisting of creating and operating abstract structures. The potential barriers to AI development concern to these two definitions. We will try to identify these barriers on the basis of the classification of cognitive operations because in our opinion it specifies the essence of generic cognitive processes adequately. We are intended to distinguish three acts of the intellect namely concept comprehension pronouncing a judgement and reasoning.
Reasoning is defined as proceeding from one proposition to another according to reliable rules of deduction. Since a simulation of human reasoning should be performed according to logical principles we use them for designing AI systems. Unfortunately in the case of a simulation of concept comprehension research results are not so impressive. The standard Aristotelian approach to concept definition which consist of giving its nearest genus and its specific difference is used in formal sciences successfully but it is not so effective in other sciences and it is usually inadequate in everyday life. There are two reasons for the difficult of applying this approach in AI firstly it is a general principle which is a troublesome for defining an effective algorithm and secondly its concepts are based on the assumption of the existence of crisp categories. Consequently a process of abstracting is treated as an intrinsic intellectual process of comprehending the heart of the matter.
The process of pronouncing a judgement is a generic cognitive process which is little understood in psychology and philosophy. In order to discuss the possibility of a simulation of this process in AI we use the Kantian taxonomy of propositions. Simulation of a generation of analytic propositions a priori is performed in AI systems. Let us recall that such propositions concern knowledge already existing in our minds. In AI systems in the case of such propositions we refer to a knowledge base directly e.g. we find the correct part of a semantic network or we derive a required proposition with the resolution method. Unfortunately we are still unable to simulate the process of generating mathematical theorems which is a fundamental process of mathematical development.
Two problems identified above which are fundamental barriers to AI development result in more specific key problems. The fact that we do not know the mechanisms of concept comprehension results in difficulty with developing satisfactory methods of automatic generation of ontologies in the area of knowledge representation and learning automatic construction of abstract models of problems in the area of problem solving and semantic analysis in Natural Language Processing.
The lack of models which describe the process of pronouncing a judgement is the main barrier in the areas of planning automatic learning social intelligence and creativity.
These barriers should not be used as an argument against the possibility of the development of intelligent systems in the future. They constitute the main challenge for research in AI.

<h2 style="text-align: left;">3. Determinants of AI Development
Most AI models have been defined on the basis of ideas which are outside computer science. Cognitive simulation semantic networks frames scripts and cognitive architectures have been developed on the basis of psychological theories. The models of standard reasoning and non-monotonic reasoning are logical theories. Genetic algorithm evolution strategies evolutionary programming genetic programming swarm intelligence and artificial immune systems are inspired by biological models. Mathematics has contributed to Bayes networks fuzzy set rough sets and standard pattern recognition. Theories of neural networks simulate models of neuroscience. Physics delivers methods based on statistical mechanics which make algorithms of problem solving and learning algorithms more efficient. It seems that only rule-based systems have been defined in computer science.
Thus the development of Artificial Intelligence treated as a research area has been influenced strongly by the theories of the scientific disciplines mentioned above. It seems that AI will be developed in a similar way in the future.
Now let us try to identify the most important AI prospects of the disciplines mentioned above.
The crucial barriers in the areas of general problem solving automatic learning Natural Language Processing planning and creativity result from our lack of psychological models of two generic cognitive processes namely concept comprehension and pronouncing a judgement. Any research result relating to these processes would be very useful as a starting point for studies into a computer simulation of these processes.
Communication between humans and AI systems and between AI systems requires much more effective NLP methods. Advanced models of syntax analysis developed in linguistics are successfully used in AI. Let us hope that adequate models of semantic analysis will be defined in linguistics in the very near future.
If advanced neuroimaging and electrophysiology techniques in neuroscience allow us to unravel the mysteries of the human brain then this will help us to construct more effective connectionist models. Models of organisms and their physiological processes and evolutionary mechanisms will be an inexhaustible source of inspiration for developing general methods of problem solving.
Further development of new logical calculi of a descriptive power that allows us to represent many aspects of the physical world would allow a broader application of reasoning methods in expert systems. Mathematics should help us to formalize models of biology psychology linguistics etc. that could be used in AI.
AI methods are very often computationally inefficient. In order to develop efficient AI methods we should use computational models that are based on mechanical statics or quantum mechanics delivered by modern physics.
New effective techniques of software and system engineering should be developed in computer science. This would allow us to construct hybrid AI systems and multi-agent systems. The algorithmization of methods which are based on models developed in various scientific disciplines is the second goal of AI research in computer science.
Hopefully philosophy will deliver modern models of theory of mind and epistemology. They play an important and inspiring role in progress in Artificial Intelligence.
Finally let us notice that AI researchers should cooperate more strongly because of the interdisciplinary nature of this research area. What is more any AI researcher should broaden his/her interests beyond his/her primary discipline. The development of a new discipline cognitive science should help us to integrate various scientific disciplines that contribute to progress in Artificial Intelligence.

<h1 style="text-align: left;">V. Conclusion
According to what we have mentioned above Artificial Intelligence deals with the attempts to model and apply the intelligence of the human mind which will determine whether a machine can think like a human by mimicking human conversation. The discipline of AI has numerous facets. Underlying all of them is the need to represent knowledge in a form that can be processed efficiently. Most parts of AI have been developed and applied in various sectors in society such as industry military medicine science biology psychology etc. Even if many companies have invested in the research of AI for many years but developing AI systems to be more perfect and efficient is still a big challenge which likely needs such a long period of time. However we are still see AI in an optimistic vision which we believe that it will play such an necessary role in the very near future.
References

1. Mariusz Flasinki (2016) Introduction to Artificial Intelligence first edition Springer International Publishing Switzerland.

2. Wolfgang Ertel (2011) Introduction to Artificial Intelligence first edition Springer London Dorderecht Heidelberg New York.

3. Stuat J. Russell and Peter Norig (2011) Artificial Intelligence: A Modern Approach third edition Tsinghua University Press.

4. Nell Dale and John Lewis (2016) Computer Science Illuminated fifth edition China Machine Press.

5. J. Glenn Brookshear (2015) Computer Science an Overview eleventh edition Posts and Telecom Press.

6. Anneli Williams (2013) Research Improve Your Reading and Referencing Skills first edition HaperCollins Publisher.

Artificail Intelligence: Application and Future Prospect Version 0
👤 Author: by ponpiseth8163com 2017-04-14 12:57:29

Reversion History