AI has developed a large number of tools to solve the most difficult problems in computer science. Some of them are listed below.
Bayesian Network:
Many problems in AI require the agent to operate with incomplete or uncertain information. AI researchers have devised a number of powerful tools to solve these problems such as the Bayesian Network, which is key computer technology for dealing with probabilistic events and to solve a problem which has uncertainty. It can also be used in various tasks, including prediction, anomaly detection, diagnostics, automated insight, reasoning and time series prediction.
The Bayesian Network can be used in medicine- machines and other computer devices assisting us in the diagnosis of the disease in order to provide better healthcare - and in semantic search - by understanding searcher intent and the contextual meaning of terms, it improves search accuracy, generating more relevant results.
Evaluating Progress:
The Turing test is a general procedure to test the intelligence of an agent and it allows almost all the major problems of AI to be tested.
However, it is a very difficult challenge and right now all agents fail. AI can also be evaluated on specific problems such as small problems in chemistry, hand-writing recognition and game-playing. Smaller problems provide more achievable goals and there is an ever-increasing number of positive results.
There is another way to measure machine intelligence which is through tests that are developed from mathematical definitions of intelligence. These tests use notions from Kolmogorov complexity and data compression. Two major advantages of mathematical definitions are their applicability to non-human intelligences and their absence of a requirement for human testers.
A derivative of the Turing test is CAPTCHA, which helps to determine that a user is an actual person and not a computer posing as a human. Unlike other tests, CAPTCHA is administered by a machine and targeted to a human. The computer asks a user to complete a simple test whose problem a computer is unable to solve. Therefore, correct solutions are deemed to be the result of a person
taking the test. A common type of CAPTCHA is the test that requires the typing of distorted letters, numbers or symbols that appear in an image undecipherable by a computer.
Search and optimization:
Since simple exhaustive search through numerous possible solutions for many problems in AI are rarely sufficient because the search space (the number of places to search) quickly grows to astronomical numbers, making the search too slow or unfinished, AI now uses “heuristics” which eliminates choices that are unlikely to lead to the goal, limiting the search to a smaller size since it provides the program with a “best guess” for the path on which the solution lies.
Another distinct type of search is based on optimization. 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. Evolutionary computation uses a form of optimization search. For example, they may begin with a population of organisms (the guesses) and then allow them to mutate and recombine, selecting only the fittest to survive each generation (refining the guesses).
Bibliography:
https://pt.wikipedia.org/wiki/Rede_bayesiana https://www.javatpoint.com/bayesian-belief-network-in-artificial-intelligence https://data-flair.training/blogs/bayesian-network-applications/ https://en.wikipedia.org/wiki/Artificial_intelligence#Tools https://www.geeksforgeeks.org/artificial-intelligence-an-introduction/ https://en.wikipedia.org/wiki/Progress_in_artificial_intelligence#Super-human
Comentários