CC BY
1APPLICATION OF ARTIFICIAL NEURAL NETWORKS IN THE
DIGITAL ECONOMY
1 2
Ayupova Diana Anatolevna, Isoqov Diyorbek Dilshod o'g'li,
3
M.Y.Doshchanova
1yulduzxon_85@mail.ru Student of TUIT named after Muhammad al-Khwarizmi,
isoqovdiyorbek742@gmail .com Student of TUIT named after Muhammad al-Khwarizmi,
"5
xodj ayevayulduzxon2021 @gmail.com Associate professor of TUIT named after Muhammad al-Khwarizmi https://doi.org/10.5281/zenodo.12633194
In the past few years, there has been an increased interest in neural networks, which are successfully used in various scientific fields - economics, medicine, technology, geology, physics. Neural networks have entered the practical application where it is necessary to solve the problems of forecasting, classification or control. For decades, scientists around the world have been working to create artificial intelligence that could match and surpass that of humans.
Thus, one of the most important in the history of the development of artificial intelligence was the victory over a person in checkers, chess, and on March 16, 2016 in go, a board game that originated in ancient China, which was considered before that, it is impossible for artificial intelligence to win, because logical thinking alone is not enough to win. Such a breakthrough was made thanks to the invention of an artificial neural network. Scientists have created this unique self-learning computer algorithm in the image and likeness of the neural network of living organisms [1].
Artificial neural network today is one of the most advanced areas in the field of artificial intelligence. John McCarthy, an American computer scientist, proposed the term "artificial intelligence" in 1956 at a conference at Dartmouth
University. In his opinion, intelligence within this science is understood only as a computational component of the ability to achieve goals.
Today it is a catch-all term covering everything from robotic process automation to real robotics. Recently, it has gained prominence, in particular, due to large amounts of data (Big Data), as well as an increase in the speed of data processing, the size and variety of data collected at a given point in time. It should be noted that robotics is now developing exponentially and for it just neural network algorithms are a powerful tool for the development of control systems, image recognition systems, audio recognition systems and other information technologies. So, an artificial neural network is a model of the work of biological neurons - brain cells, which is represented by a set of mathematical instructions written in the form of a program code.
However, to call a neural network a classical algorithm is not entirely correct. It is much more complicated. A huge number of processes in it do not occur linearly, but in parallel at one point in time, i.e. just like in our brain. In an artificial neural network, each neuron is represented as a processor. It has channels for receiving and outputting a signal. At the input, each signal passes through certain connections that mimic the synaptic activity of biological neurons, i.e. their ability to communicate information to each other. Each artificial neuron can work with only one unit of incoming information and performs the simplest function. However, as soon as they are networked, they can already cope with the most complex tasks that are beyond the power of conventional programming methods.
However, the main feature of the game is not that it is impossible to calculate all the options of the game, but that in order to win you need to connect intuition, a quality inherent in a person. The principle of operation of artificial neural networks is based on the so-called machine learning. Machine learning is a class of artificial intelligence methods, the characteristic feature of which is not the direct solution of a problem, but learning in the process of applying solutions to many similar problems [2]. There are two types of training: learning by precedents, or inductive learning, is based on the identification of empirical patterns in the
data, deductive learning involves the formalization of experts' knowledge and their transfer to a computer in the form of a knowledge base. Many inductive learning methods have been developed as an alternative to classical statistical approaches and are closely related to information extraction methods and data mining methods. In this case, to develop intuition in the AlphaGo program, machine-learning technologies were used. Like humans, a neural network learns by trial and error. So, based on errors, the system corrects its work itself. This process is performed until the neural network begins to produce a consistently correct result.
The training period depends on the complexity of the task itself and the amount of data that the neural network must remember. So, it can take months to train a neural network. For example, the AlphaGo program, which actually consists of several neural networks at once, was trained to play Go for six months on 50 GPUs working in parallel or distributed using the Google Cloud platform, whose servers are located in the United States. It should be noted that a feature of distributed multiprocessor computing systems, in contrast to local supercomputers, is the possibility of unlimited increase in performance due to scaling.
The neural network was trained to predict the next move based on a large number of games played. At this time, as another neural network was trained to guess who will win. Those. intelligence lies in the fact that no one explained to the program how to play correctly. The machine has learned to do it itself. The main world leaders in the field of neural network technologies should be considered Google and its divisions (DeepMind, AlphaGo, and Google Brain), Microsoft (Microsoft Research), IBM, Facebook (Facebook AI Research division) and others. So, another no less significant success in the field of artificial neural networks is the IBM supercomputer Watson.
Thus, the operation of such an information system allows real-time assistance to the operator to optimize the production management process. It should be noted that, in contrast to human information system does not get tired and provides stable recommendations to ensure the production process in the best mode. In addition, the information system helps the furnace operator to reduce the
operating time under current by 10%, reduce the cost by more than 3%, and ensure stable quality. As noted above, such systems can be used in metallurgical, chemical, oil refining and power generation industries. The introduction and use of the information system does not require a deep modernization of production, which is a significant advantage [3,4].
The neural network approach is widely used in medicine. A team of researchers from the University of Nottingham has developed four machine-learning algorithms to assess patients' risk of cardiovascular disease. Data from 378,000 British patients were used for training. Trained artificial intelligence determined the risk of cardiac diseases more effectively than real doctors did. Artificial intelligence has significantly improved recommendation mechanisms in online stores and services. Algorithms based on machine learning analyze your behavior on the site and compare it with millions of other users. All in order to determine which product you are most likely to buy. In agriculture, in May 2016, Cognitive Technologies began testing an unmanned tractor equipped with computer vision thanks to satellite sensors. Unmanned vehicles are supposed to be used for harvesting, weeding, spraying plants with herbicides and other processes. Note that unmanned vehicles, a concept that most large corporations, as well as technology companies (Google, Uber, Yandex and others) and startups, are working on, relies on neural networks in their work. Artificial intelligence is responsible for recognizing surrounding objects - be it another car, pedestrian or other obstacle. A recent IBM survey found that 74% of automotive industry leaders expect smart cars to be on the road by 2025. Such IoT-integrated vehicles will collect information about passenger preferences and automatically adjust the interior temperature, radio volume, seat position and other parameters. In addition to piloting, the system will also inform about emerging problems and traffic situations. In logistics and manufacturing, automatic unmanned loaders are widely used. The use of artificial intelligence in marketing activities allows you to collect and quickly analyze information about thousands of users to promote goods and services.
In the field of personnel management, artificial intelligence is already being used to process resumes, conduct interviews, and also monitor the actions of employees to prevent fraud. In addition, machine learning helps to recognize potential cases of fraud in various areas of life. PayPal - as part of the fight against money laundering, uses a similar tool, for example, the company compares millions of transactions and detects suspicious ones among them. As a result, PayPal's fraudulent transactions are at an all-time low of 0.32%, while the standard in the financial sector is 1.32%. The work of the police and firefighters already involves the use of artificial intelligence. The cameras installed, for example, in London, not only record the fact of a crime, but also independently prepare documents for sending to the prosecutor's office.
The use of artificial intelligence in the education industry makes it possible to automate the testing of tests, as well as to develop perfect methods for transferring knowledge. For example, the robot Alantim, an employee of the Moscow Institute of Technology, lectures students on robotics and conducts a large-scale study of the interaction between people and smart machines. This is the first robot to be accepted into an official position at the institute. Perm developers from Promobot developed Alantim. Alantim is able to maintain a meaningful conversation, answer questions. The robot remembers up to 1,000 faces, uses 400,000 speech modules in a conversation. The maximum operating time of the robot is 8 hours without recharging, the speed of movement is 5 km/h. Alantim saves and systematizes all the information received, and then sends it to the owners in the form of reports. State-of-the-art technology and automation are increasingly penetrating our daily lives. Naturally, our home is no exception.
A smart home is one of the priority areas for the development of the Internet of things (IoT), in which a huge number of companies operate. The intelligent system "Smart Home" allows you to combine all communications and perform control and management using artificial intelligence, programmable and customizable to all the needs and wishes of the owner. So, the "Smart" house regulates the temperature in the room, starts the necessary equipment and performs
dozens of other useful functions. Summing up, it should be noted that in the future the use of artificial intelligence will only expand. There is a growing demand for high-level designers of automated systems and programmers. These professions are already called the professions of the future. However, it should be noted that there are problems with the use and implementation of artificial intelligence.
The use of artificial intelligence has led to the gradual destruction of a number of professions in the labor market. Over time, cashiers, drivers, accountants, estimators will disappear - machines already perform these and other operations. This will lead to a sharp increase in unemployment, as well as increased demands from employers. New qualified personnel will be required to work with automated systems, and this will lead to a revision of the education industry.
References:
1. Bondarev V.Yu., Sorokin A.S., Krotova Ye.L. Iskusstvennaya neyronnaya set kak sredstvo i metod statisticheskoy obrabotki dannyx // Vestnik UrFO. Bezopasnost v informatsionnoy sfere. 2016. -№ 2 (20). - S. 19-22.
2. Tixonov A.A. Bolshiye dannye i glubokoye mashinnoye obucheniye v iskusstvennyx neyronnyx setyax // Nauka i obrazovaniye segodnya. 2018. - № 6 (29). - S. 35-38.
3. Doshanova M., Dilmurodov T., Karimov M. PROBLEMS OF ADAPTIVE NEURO-FUZZY MODELING OF COMPLEX OBJECTS //International Scientific and Practical Conference on Algorithms and Current Problems of Programming. - 2023.
4. Nodirbek Y., Gulyamov S., Doshchanova M. Neuro-fuzzy modeling for predictive control systems with complex technological processes and production //Chemical Technology, Control and Management. - 2020. - Т. 2020. - №. 1. - С. 73-83.
