In this article, we would like to discuss in detail the concept of artificial intelligence methods. To explain the essence of this concept in clear language, we must first understand the concept of “method” itself.
What is an artificial intelligence method?
The term “method” has many definitions. Since the field of artificial intelligence mainly implies knowledge in the field of mathematics, programming, betting on Safe Casino, and information technology, the method, in our case, is a path of knowledge or a way of knowing any subject area, a way to achieve a goal. And the method of artificial intelligence is a way, or in fact, an algorithm for solving any problem.
There are a large number of areas of artificial intelligence development. Within these areas, there are various methods that can be used individually or in groups to solve problems facing science, industry, economics, medicine, and other areas.
Classification of artificial intelligence (AI) methods
There are different opinions on how to classify AI methods. We propose the following classification, which consists of five points:
Artificial neural networks
- Fuzzy logic (fuzzy sets and soft computing)
- Knowledge-based systems (expert systems)
- Evolutionary modeling (genetic algorithms, multi-agent systems)
- Machine Learning (Data Mining and data analysis, searching for patterns in data warehouses)
Now, let’s explain in simple terms what each method is.
Artificial neural networks
An artificial neural network is primarily a mathematical apparatus, although sometimes elements of logic are found in various neural network paradigms.
A neural network is a mathematical model whose prototype is the central nervous system of a human or animal.
This AI method is used in problems of pattern recognition, forecasting, classification, clustering, and optimization.
Fuzzy logic, fuzzy sets, and soft computing
Fuzzy logic, fuzzy set theory, fuzzy reasoning, and soft computing are all close or closely related concepts related to a higher level of the central nervous system than artificial neural networks. Fuzzy logic methods are used in expert systems, object control systems.
Fuzzy logic is more associated with a qualitative assessment of the analyzed processes and phenomena and decision-making based on this qualitative assessment.
Evolutionary or multi-agent modeling
This group of methods considers the concept of not individual, but collective intelligence.
Evolutionary modeling is appropriate to use when the solution search space is so large and complex that traditional and simpler methods are simply unable to perform a global search for a solution, or are capable, but this would require an unacceptably long time.
Expert systems. Decision support
An expert system is an artificial analog of a decision maker, or at least an expert consultant in a subject area.
The structure and logical-mathematical apparatus of an expert system are determined, foremost, by its purpose and subject area. The solutions offered by the system can be developed using various inference mechanisms. The closest analog to a human inference mechanism is the apparatus of fuzzy logic and fuzzy set theory.
Machine Learning, Data Mining, Data Science
Machine Learning is a whole class of artificial intelligence methods. All of them involve solving problems not directly, but through preliminary training both before and during the decision-making process.
Data mining. This term was introduced by Grigory Pyatetsky-Shapiro in 1989
In essence, this is a collective name that is used to denote a whole group of methods for detecting certain patterns in the total volume of data that can be obtained in various areas of human activity. For example, Data Mining methods can be used for Big Data accumulated in retail sales to confirm any hypotheses and make management decisions.
Conclusions
We have considered 5 main groups of artificial intelligence methods according to our classification and provided brief explanations regarding each of them.
In other posts, we will consider each method in more detail.
