Artificial Intelligence

Artificial Intelligence Ayushi Pradhan

Artificial Intelligence A property of machines that, if achieved, mimics human thought processes. Many researchers in artificial intelligence consider the abilities of perception, learning, reasoning, and decision making as essential to claims of machines possessing artificial intelligence.

Computational Intelligence While some techniques within computational intelligence are often counted as artificial intelligence techniques (e.g., genetic algorithms, or neural networks) there is a clear difference between these techniques and traditional, logic based artificial intelligence techniques. In general, typical artificial intelligence techniques are top-to-bottom, where the structure of models, solutions, etc. is imposed from above. Computational intelligence techniques are generally bottom-up, where order and structure emerges from an unstructured beginning. The areas covered by the term computational intelligence are also known under the name soft computing.

Requirementsof Artificial Intelligence Systems dependent on artificial intelligence would normally require more processing capabilities than normal systems.

Hardware While speed and memory of an AI system is important intelligent systems quite often require a range of sensors to receive input data from the environment. Output devices will include the normal peripheral devices such as printers and monitors but may also include a range of actuators or speech synthesis devices. Robotics are one application of intelligent systems and are used in CAM systems.

Software AI is dependent not only on sufficient hardware but also on the software to run the hardware and to synthesize the data received. Once the data has been received and processed the AI system needs to make an intelligent response. To create this software non-procedural languages are often used. These include languages such as LISP and PROLOG. Both of these languages will actually allow the system to learn and modify its responses to its environment.

Specific Requirements Models and Simulations: Models and simulations may require high graphics capabilities (3D photorealism) and be capable of processing high end mathematical models which can be very CPU intensive. Machines like this will require: a fast CPU large amounts of RAM A good graphics card large storage capacity (i.e. large hard drive) may require specialized input output devices will require specialized software and may require an AI language such as PROLOG or LISP

Neural Networks: A wide range of software is available for developing neural networks, some can even be used as plugins for spreadsheets. The requirements will vary according to the use but a neural network would normally store large amounts of data and have the data linked in a configuration which would require a large storage capacity hard drive, fast access to the data would also be an advantage. If the network uses graphics then a fast graphics adapter would also be required. Expert Systems: Expert systems are specialised information systems which would normally require a large storage capacity and fast processing capability. Many of these will be used via the internet so a fast connection would also be necessary

AI Techniques 1 Searching: Search plays a major role in solving many (AI) problems. Search is a universal problem-solving mechanism in AI. In many problems, sequence of steps required to solve is not known in advance but must be determined by systematic trial-and-error exploration of alternatives. The problems that are addressed by AI search algorithms fall into three general classes:

1. SINGLE-AGENT PATH-FINDING PROBLEMS Classic examples in the AI literature of path-finding problems are sliding-title puzzles, Rubik’s Cube and theorem proving. The single-title puzzles are common test beds for research in AI search algorithms as they are very simple to represent and manipulate. Real-world problems include the traveling salesman problem, vehicle navigation, and the wiring of VLSI circuits. In each case, the task is to find a sequence of operations that map an initial state to a goal state.

2. TWO-PLAYERS GAMES Two-players games are two-player perfect information games. Chess, checkers, and Othello are some of the two-player games. 3. CONSTRAINT SATISFACTION PROBLEMS Eight Queens problem is the best example. The task is to place eight queens on an 8*8 chessboard such that no two queens are on the same row, column or diagonal. Real-world examples of constraint satisfaction problems are planning and scheduling applications.

AI Techniques 2 Pattern Recognition: A branch of artificial intelligence concerned with the classification or description of observations. Pattern recognition aims to classify data (patterns) based on either a priori knowledge or on statistical information extracted from the patterns. The patterns to be classified are usually groups of measurements or observations, defining points in an appropriate multidimensional space. It is the research area that studies the operation and design of systems that recognize patterns in data. It encloses sub disciplines like discriminant analysis, feature extraction, error estimation, cluster analysis (together sometimes called statistical pattern recognition), grammatical inference and parsing (sometimes called syntactical pattern recognition). Important application areas are image analysis, character recognition, speech analysis, man and machine diagnostics, person identification and industrial inspection.

AI Techniques 3 Heuristics: Heuristic search is an AIsearch technique that employs heuristic for its moves. Heuristic is a rule of thumb that probably leads to a solution. Heuristics play a major role in search strategies because of exponential nature of the most problems. Heuristics help to reduce the number of alternatives from an exponential number to a polynomial number. In Artificial Intelligence, heuristic search has a general meaning, and a more specialized technical meaning. In a general sense, the term heuristic is used for any advice that is often effective, but is not guaranteed to work in every case. Within the heuristic search architecture, however, the term heuristic usually refers to the special case of a heuristic evaluation function.

HEURISTIC INFORMATION In order to solve larger problems, domain-specific knowledge must be added to improve search efficiency. Information about the problem include the nature of states, cost of transforming from one state to another, and characteristics of the goals. This information can often be expressed in the form of heuristic evaluation function, say f( n, g), a function of the nodes n and/or the goals g.

AI Techniques 4 Machine Learning: To solve problems computers require intelligence. Learning is central to intelligence. As intelligence requires knowledge, it is necessary for the computers to acquire knowledge. Machine learning serves this purpose. Machine learning refers to a system capable of acquiring and integrating the knowledge automatically. The capability of the systems to learn from experience, training, analytical observation, and other means, results in a system that can continuously self-improve and thereby exhibit efficiency and effectiveness. A machine learning system usually starts with some knowledge and a corresponding knowledge organization so that it can interpret, analyze, and test the knowledge acquired.

Applications of AI Game playing: Youcan 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 brute force computation--looking at hundreds of thousands of positions. To beat a world champion by brute force and known reliable heuristics requires being able to look at 200 million positions per second. Speech Recognition: In the 1990s, computer speech recognition reached a practical level for limited purposes. Thus United Airlines has replaced its keyboard tree for flight information by a system using speech recognition of flight numbers and city names. It is quite 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. Understanding Natural Language: Just getting a sequence of words into a computer is not enough. 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.

Computer Vision: The world is composed of three-dimensional objects, but the inputs to the human eye and computers' TV 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. 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 whether 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 (e.g., about whether there have been previous credit card frauds at this establishment).

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 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, hospitals, death, recovery, and events occurring in time. Its interactions depended on a single patient being considered. Since the experts consulted by the knowledge engineers knew about patients, doctors, death, recovery, etc., it is clear that the knowledge engineers forced what the experts told them into a predetermined framework. In the present state of AI, this has to be true. The usefulness of current expert systems depends on their users having common sense.

Some Important Terms & Their Definitions

Turing test A test for deciding whether a computer is intelligent, proposed in 1950 by the mathematician Alan Turing. He preferred to consider if machines can be intelligent as opposed to whether "they can think". In a Turing test, a human converse in writing with an unseen person or machine. If the human cannot distinguish between an unseen human and an unseen machine (computer) then the machine is said to have passed the test and is intelligent.

In the basic Turing Test, there are three terminal s. Two of the terminals are operated by humans, and the third terminal is operated by a computer. Each terminal is physically separated from the other two. One human is designated as the questioner. The other human and the computer are designated the respondents. The questioner interrogates both the human respondent and the computer according to a specified format, within a certain subject area and context, and for a preset length of time (such as 10 minutes). After the specified time, the questioner tries to decide which terminal is operated by the human respondent, and which terminal is operated by the computer. The test is repeated many times. If the questioner makes the correct determination in half of the test runs or less, the computer is considered to have artificial intelligence, because the questioner regards it as "just as human" as the human respondent.

CAPTCHA A CAPTCHA (Completely Automated Public Turing test to tell Computers and Humans Apart) is a challenge-response system test designed to differentiate humans from automated programs. A CAPTCHA differentiates between human and bot by setting some task that is easy for most humans to perform but is more difficult and time-consuming for current bots to complete. CAPTCHAs are often used to stop bots and other automated programs from using blogs to affect search engine rankings, signing up for e-mail accounts to send out spam or take part in on-line polls.

Frequently, a CAPTCHA features an image file of slightly distorted alphanumeric characters. A human can usually read the characters in the image without too much difficulty. A bot program is able to recognize that the content contains an image , but it has no idea what the image is. To accommodate the visually-impaired, some CAPTCHAs use audio files. In such a system, the human listens to a series of letters or short words and types what he hears to prove he is not a bot.

Natural Language Language that is close to the everyday speech of human beings. Machine Learning An set of techniques in artificial intelligence that makes it possible for a machine's performance to improve based on feedback from previous performances.

Neural network Neural nets, networks of distributed, parallel processing computer systems based on the structure and functioning of the human brain Pattern Recognition The ability to identify the underlying patterns in input data.

Heuristics Generalization or rule of thumb that directs someone's attention in a way that helps them learn or discover e.g. if you are having trouble understanding a problem, try drawing a diagram of it. Programming in a heuristic way is a major skill in building artificial intelligence. Instead of rigorous mathematical algorithms, it involves general strategies. Good anti-virus software uses heuristic detection - spotting patterns of behavior that "aren't quite right" instead of just relying on recognizing virus code

Fuzzy Logic A type of logic used to process conditions that are only partly true or false. It deals with probabilities instead of the more classic Boolean logic which can only deal with TRUE and FALSE. Practical applications in computer controlled systems include the control of fuel and air mixtures in internal combustion engines, the proportional slowing of the speed of objects as they approach a given state or target, the heating and cooling of objects or spaces to prevent overheating and the mixing of two or more ingredients to achieve a defined final condition (especially when the components and their properties are constantly changing). Fuzzy logic uses weighted algorithms in computer programs to simulate human thought or life-like responses to external conditions.

Fuzzy logic in it's simplest terms expands the dichotomy of true or not true to include a range of answers in between. The usual example is say instead of being black or white, fuzziness allows for shades of gray. Since fuzzy logic allows this extra bandwidth in fuzzy answers, fuzzy rules used in programming can cover a much broader area. A fuzzy rule such as "When it rains, you get wet"*** can cover a lot of ground. It would be able to several instances of itself such as "when it rains a lot, you get wet a lot" or "when it rains a little, you get wet a little". Rules like this are beautiful because they are human rules. They are a much better model of how we think. It is not often that questions that arise in life have bivalent answers. There are a few that do such as "Are you married?". Other questions such as "Do you like your job?", would tend to elicit a range of a response falling somewhere between yes and no. How exactly is a fuzzy rule able to cover so much ground? By the use of a patches. A fuzzy rule will define a fuzzy patch. Say for example that you would like to use fuzzy logic to control an air conditioner. You could define a fuzzy set for the temperature range as COLD, COOL, JUST RIGHT, WARM and HOT. A system could be composed of a few sloppy rules with wide patches, or many precise rules with narrow patches. Perhaps the air conditioner system is representative of other real systems. That is, an optimal solution involves some wide sloppy rules, and some precise ones.

The fuzzy rules that would go with the air conditioner system would be: Rule 1: If the temperature is cold, them motor speed stops.Rule 2: If the temperature is cool, the motor speed slows.Rule 3: If the temperature is just right, the motor speed is medium.Rule 4: If the temperature is warm, the motor speed is fast.Rule 5: If the temperature is hot, the motor speed blasts.This fuzzy system works well because the patches will cover lines that correspond to relations between temperature and motor speed if they are non-linear and squiggle. In contrast a similar bivalent system might be built of many specific rules such as if temperature is 60 degrees than the motor speed is 50. Patches developed from rules like this would only be points, and the system developed from it would only define a collection of points-- not a terrific model. Therein lies a greatness of fuzziness.

Agents- Help or Hindrance? A software agent is a program that acts on behalf of the user or another program and therefore has some authority to change its behavior depending on circumstance. It must therefore have some degree of artificial intelligence. They might ask questions and respond to commands in different ways according to a user's work patterns, or use reasoning to set goals for a user.

A few Issues The increasing use of robotics, artificial intelligence (AI) and expert systems raises a range of ethical issues. For example, at which point should humans hand over key decision-making to a computer? Should robots have the same rights as humans? What social impacts might arise with the replacement of human workers or the creation of smart weapons?

Bibliography http://www-formal.stanford.edu/jmc/whatisai/node4.html http://www.digitalist.org/Itgs/theory/12_ai_expsys/12_ai_expsys.htm http://www.cs.vu.nl/ci/ http://searchsecurity.techtarget.com/definition/CAPTCHA http://intelligence.worldofcomputing.net/ai-search/ai-search-techniques.html http://aaai.org/AITopics/PatternRecognition http://intelligence.worldofcomputing.net/machine-learning/machine-learning-overview.html http://mark.stosberg.com/Tech/fuzzy/role_in_ai.html http://www.bettscomputers.com/requirementsofai.htm

Artificial Intelligence