How will neural networks help the American army?
Neural networks and programs for the study, creation and implementation of artificial intelligence have been popular in recent years. These developments are of interest to the military in all countries of the world. Since 2017, the country that spends the most money on military artificial intelligence systems is the United States. Considering the size of the US military budget, which is many times higher than the military budgets of other countries of the world, this is nothing special.
A 2018 study by the analyst company MarketsandMarkets of the global market for AI technologies used for military purposes showed that back in 2017, the world spent 6,26 billion dollars for these purposes (this is more than the annual budget of a country like Nepal). The figure was obtained by adding up the sales value of the corresponding software, services and equipment. According to analysts, by 2025 the volume of the world market for military artificial intelligence technologies will reach $ 18,82 billion (between the annual budget of Bulgaria and Uruguay).
These technologies are of interest to the military of different countries. In Russia, the Ministry of Defense plans to create a separate department for work with artificial intelligence. The timing of the creation of such a department became known on May 31, 2021. According to Mikhail Osyko, who is a member of the board of the Military-Industrial Commission under the Russian government, the new structure will have its own budget. The Russian military plans to complete the formation of the department by December 1, 2021.
Programmable Uniform Fiber
One of the developments in the field of neural networks is a programmable fiber that will transmit data about the state of the fighter. The development of this material is being carried out by American scientists from the Institute for Soldier Nanotechnologies, which is part of the structure of the famous Massachusetts Institute of Technology. It is assumed that the new programmable fiber, which can be used in the tailoring of military uniforms, will be able to provide an extensive set of biometric data to monitor the condition of a soldier.
At the beginning of September 2021, journalists from the American edition of C4ISRNET (The Journal of Net-Centric Warfare) wrote about these developments. This is a specialized publication that covers for federal officials and representatives of the defense industry the problems and trends in the transformation of military technologies, as well as network technologies that can be used for military purposes.
As noted by American journalists, a special programmable fiber, which is being worked on at the Institute of Military Nanotechnology, will be able to capture, analyze, store and issue data on the activity of fighters. It should be noted that this technology is at the initial stage of development - at the stage of conceptual tests.
Tests should demonstrate the feasibility of creating a fiber that will store information. As part of the experiments, American scientists were able to load a film onto cloth in order to clearly demonstrate the possibility of storing large amounts of data. This work is supervised by the US Army Research Lab (ARL for short).
As a development of this technology, scientists want to implement algorithms into the fiber that will help to analyze all the collected information faster. According to ARL, the scientific breakthrough is the creation of an experimental fiber that has a whole neural network of 1650 connections. The network created by the engineers for 270 minutes successfully collected data on the external body temperature of the soldier. In laboratory conditions, thanks to the capabilities of AI, on the basis of temperature data, the fiber almost always determined the type of activity of a soldier. The accuracy of the results obtained was 96 percent.
In the distant future, scientists plan to develop these technologies in different directions. In particular, work will be carried out to ensure that the fiber has the ability to store and generate energy for AI analysis or for the operation of communication systems and sensor devices from a military outfit.
James Burgess, program director of the research laboratory at the Institute of Military Nanotechnology, believes that the development of these technologies will save the lives of military personnel. Thanks to the special fabric and the presence of wearable computers, the uniform will be able to transmit information about the condition of the fighters when it is no longer possible to contact them in the traditional way. Without seeing the soldier, the medics and the command will be able to obtain information about his condition: whether he is breathing, what is the soldier's pulse and body temperature.
At the same time, American developers do not hide that they are at the very beginning of the journey. All smart fiber developments are in the early stages of research. They can have some kind of impact on the battlefield only by 2050. According to Burgess, the developments of scientists from Massachusetts will also be able to promptly provide soldiers with information about their physical condition.
With the help of signals a soldier can be warned of dehydration, overheating. According to Burgess, it is very important that in combat conditions, including in an adrenaline rush, servicemen always know what is happening to them, and also correctly assess their physiological state.
Neural networks will help commanders
Another American project that scientists are working on in collaboration with the US Army Research Laboratory is to create a working tool for military commanders that can quantify uncertainties. An algorithm using a neural network will work with large amounts of data and analyze them. AI-powered neural networks will form a confidence rating. Ultimately, this will allow commanders to be more confident in their decisions.
In a combat situation, this is especially important, since concentration of attention can be lost, while, being in a stressful situation, people can lose sight of certain details and indicators, or, conversely, overestimate some values. Neural networks and AI, processing large amounts of data, will have to reduce the risk of possible errors in decision-making.
At the same time, the neural network will help assess how inaccurate artificial intelligence systems can be. The criterion for evaluation is important because people need to realize that even AI is imperfect. At the same time, the commander will always be able to imagine how correct the calculations made by the AI are in order to operate with this data and make their own decisions based on them.
At the same time, many factors can always influence the inaccuracy of the data obtained in battle and the conclusions that the AI draws on their basis. First, the data can be manipulated, it can be done by the enemy. Secondly, the received data may contain random "noise" that introduces uncertainty or violates established algorithms. Thirdly, uncertainty can arise from failed sensors damaged by the enemy.
In order to quantify the assessment of possible uncertainty, American researchers are working on a classification of possible sources of uncertainty encountered in military networks. They are helped in this by neural networks, which are excellent for processing large amounts of data, establishing communications and communication.
Information