17 february 2021
Results of 2020: the most interesting AI solutions and projects

Results of 2020: the most interesting AI solutions and projects

In recent years, the IT market has seen a real boom in artificial intelligence solutions. It is not surprising: modern computing and neural network technologies have reached a level that allows AI systems to solve very complex practical tasks for humans, and developers to create innovative applications and services that demonstrate the limitless potential of electronic intelligence.

The sense of smell

One of the brightest examples of intensive development of artificial intelligence technology was the AI-complex created by specialists from Intel Labs and Cornell University, which is able to distinguish odors and imitate the human olfactory nervous system work. The development was based on Intel Loihi neuromorphic processors, which combine learning, training and decision-making processes in a single chip and allow the system to be autonomous and “smart” without connecting to the database. In the course of experiments, the complex designed and equipped with chemical sensors demonstrated high efficiency in detecting odors of hazardous substances in the air, even in strong interference conditions. Such solutions, Intel believes, will help in the development of robotics, when robots will be able to sort products by themselves, guided by the smell, will push the environmental monitoring systems development, will lead to improved labor safety in manufacturing and in general will give impetus to the silicon processors cognitive abilities development.

Diseases will recede

Developers of AI systems made significant progress in the field of medicine in 2020. For example, DeepMind, a company owned by Alphabet (Google), announced a significant breakthrough in the prediction of protein folding. The problem of predicting protein folding is considered to be one of the 125 most important problems of our time, and one of the greatest problems of biology for the last 50 years. The fact is that proteins are assembled from linear sequences of amino acids, which after synthesis take a unique spatial form, and there are a huge number of such forms. Only 0.1% of the hundreds of millions of proteins (combinations of amino acids) have been studied so far, whose spatial structure is also well known. Unknown proteins, as well as compounds whose properties have not yet been confirmed experimentally, scientists are trying to predict with the help of computers. But no one till now has been able to calculate with a sufficient degree of accuracy, which 3D-form of a given set and sequences of amino acids will take a protein. DeepMind claims to have found the key to solving this problem. If this is true, we can expect a breakthrough in the discovery of new medicines and vaccines, and also the understanding of the origin and course of many diseases.

The power of thought

Mind-reading is still the part of science fiction films and books. However, science and technology do not stand still, and there are enough reasons to believe that this kind of technology will become a reality in the future. A group of scientists from the University of California, San Francisco, has advanced one step forward in this direction. They experimentally proved the possibility of recognition of nerve signals in the human brain and translating them into comprehensible words using recurrent neural network and brain-implanted electrodes. Patients with epilepsy were involved in the experiment. The electrodes were implanted to fight the neurological disease and track seizures. It so happened that some of the electrodes were in the areas of the brain in which words are selected, expressions are composed and feedback is carried out with the parts of the brain that perceive a person’s own speech. The patients with epilepsy were asked to pronounce several sentences with a limited set of words mentally and then aloud. Signals from sensors implanted in the brain were recorded at the same time. The obtained data were transferred to a neural network for training, and the intermediate result was given to another AI-network for analysis. The probability of misidentifying words was only 3 percent. Impressive result!

From classical to rock

OpenAI programmers have found an interesting use for artificial intelligence in developing Jukebox, an artificial intelligence that composes music with meaningful lyrics and vocals. They used many excerpts from songs in a variety of genres, from rock, jazz, and blues to hip-hop with country and classical pieces to train the system’s neural network. This approach allowed the OpenAI team to expand the capabilities of the project and achieve the effect of imitating the musical compositions of the artists whose tracks it was trained on. For example, Jukebox can compose music in country singer Johnny Cash, rapper Drake, and even the Russian pop band Tatu style. The artificial intelligence takes about 9 hours and a huge amount of computing resources to create one minute of music with vocal parts. For this reason, the company can not provide public access to its AI-system. But the developers published the results of Jukebox.

Drawing Lessons

Electronic intelligence has found application in the visual arts. In the middle of 2020, it became known about the creation of a machine learning system Timecraft by specialists of the Artificial Intelligence Laboratory of the Massachusetts Institute of Technology. It allows to recreate the process of painting and applying strokes to the works of famous artists such as Monet, Vincent Van Gogh or Salvador Dali. It is reported that the neural network was first trained on two hundred video clips of accelerated filming techniques for painting real digital and watercolor paintings. After that, the researchers created a convolutional neural network, which is designed to “deconstruct” the artwork based on their knowledge of the process of creating paintings. As a result, Timecraft was able to show higher efficiency than existing similar projects in more than 90% of cases. Not a bad result. Aside from virtual history lessons, the Timecraft AI system can be useful to illustrate common painting techniques and techniques for beginners.

Mind games

Artificial intelligence has found many applications in other spheres of human activity. For example, NVIDIA has used AI to recreate the gameplay of the famous arcade video game Pac-Man using the GameGAN neural network. Artificial intelligence needed only 4 days to solve this problem. The company trained the neural network using 50 thousand game sessions in Pac-Man. Then it was given the task to recreate the entire game she had seen, starting from static walls and dots and ending with moving ghosts and Pacman himself. The game was trained and recreated using a quartet of NVIDIA Quadro GP100 graphics accelerators. The most interesting thing is that GameGAN did not provide access to the original game code or its engine. All training boiled down to the fact that one neural network watched how another neural network played in Pac-Man. “A programmer needs to come up with and write down rules for the behavior and interaction of all available agents within the game to create a game like Pac-Man. This is a very painstaking job. GameGAN can make this easier. A neural network is capable of learning new rules through observation. Ideally, algorithms like GameGAN could be trained to generate procedural rules for the game you want to create,” NVIDIA researchers explain. They also underline that their development can be used not only in the gaming industry, but also in other areas in the future.

Vision of the future

The past 2020 can be safely called the year of the brightest achievements of artificial intelligence, which will continue its intensive development, despite the coronavirus pandemic and the difficult economic situation in the world. According to analysts from the International Data Corporation (IDC), in the past year, global costs in this area amounted to approximately US $ 156.5 billion. By 2024, the market volume will double and exceed $ 300 billion. At the same time, software will remain the largest segment in the industry. In second place in terms of costs will be various AI-services. The rest will come from hardware solutions. In the distant future, artificial intelligence will affect almost all spheres of human activity. The AI ​​market has a large reserve for the future and good prospects for development, and therefore there is every reason to believe that the dynamics of its growth will be higher than the expectations and forecasts of experts.

Subscribe to our newsletter

By clicking the button, I accept the terms of the Offer for the use of the site and agree with privacy policy