ITMO University Digest: A Selection of Resources on Artificial Intelligence

What is machine learning really? The whole truth about the novelty of the idea, aspects of modern success, errors of neural systems and changes in technology.

How drawing parallels between human thinking and holography and human thinking and artificial intelligence determines the direction of research. An example of a holographic implementation of linguistic modeling and prediction.

The interactive online book by Michael Nielsen gives an idea of ​​the basic concepts of neural networks and deep learning and modern approaches to understanding them. If you wish, you can write code for a simple neural network that can recognize handwritten numbers.

On the work in the field of machine learning, priority areas of development, the future of technology. And about solving specific practical problems - which algorithms and languages ​​are better to use.

All about the problem of generalization: the subsections of artificial intelligence and the ingredients needed to create AI, possible difficulties, methods of working with distributed representations and deep compositions.

Detailed history of artificial intelligence, written by Niels J. Nilsson. Everything you need to know about AI from the beginnings of logic and philosophy, the beginning of research in the 50s to modern concepts and technologies.

The AI ​​market is growing steadily, the noise around the phrase "artificial intelligence" is rising more and more, but what is really behind all this? About the areas in which the breakthrough occurred this year, why corporations are buying startups, and what good things AI helps to create.

Alternative scenarios: artificial intelligence will save us from incurable diseases, the planet from global warming. Or leave people out of work. Researchers offer the most plausible scenarios.

There are more and more organizations studying the effect of artificial intelligence on modern reality, but the main subject of research should be the effect of AI on humans. An article about the possible consequences of the widespread introduction of AI, which scare scientists.

Fantastic scenarios to capture the planet "evil AI" is unlikely to become a reality. But we should not forget about moral values, first of all, our own, because cars are a reflection of people.

Perhaps the time has come to add the proverb: “God created man in his own image and likeness, and man repaid in the same way”, and reread Nietzsche, who, it turns out, did not write at all about superman, but about super-machine. A little bit of reasoning about what kind of “God” AI will be.

Artificial intelligence learns very quickly and a person, or even a whole team, cannot constantly monitor his decisions. What should be taught to the virtual guardian of order and how to organize its work?

The book of the Swedish philosopher and professor at Oxford University Nick Bostrom about what artificial intelligence will do with humanity when it surpasses us in intellectual development. Or the "explosion" can be made controlled?

In his book, Bill Hibbard talks about how to avoid incorrect tasks and how knowledge of human values ​​will help set up basic AI settings so that cars keep a friendly attitude towards people.

About how the retail sector has changed, what information artificial intelligence collects about customers, how it helps sellers and makes shopping easier for buyers.

Thanks to voice recognition, it will soon be possible to make purchases right behind the wheel. About voice search, which increases customer loyalty and increases sales.

Stop reading chat bots, it's time to create them! A selection of services for creating a "layout" of the idea and the chat bot itself.

How the company is trying to make amends after the "unsuccessful" S7 series and that not only smartphones, but also refrigerators will soon become smart.

Everyone is familiar with the situation when attempts to schedule a meeting, taken in the general chat, end in failure. Soon Siri will be able to choose their own time, given the schedule of each of the participants in the chat. Read about the other useful features of the voice assistant in the article.

The article describes several neural network architectures used to analyze textual information, and neural network methods used to solve the main set of tasks for Data Mining of texts.

About how artificial intelligence has learned almost unmistakably to recognize speech only by the lips and where to apply such technology in practice. And also about what approaches scientists used to teach AI.

Kohonen's neural networks help to organize associative data access, which lies in the field of graphic associations. What new principles bring neural networks into the associative search technology in an array of unordered information?

Creating a global artificial intelligence using the semantic web and ways to implement solutions. Knowledge base of how much AI is needed? Comparison with natural intelligence.

Ray Kurzweil sure to create artificial intelligence, you must first understand your own. There are no recipes for creating powerful AI, but there are interesting parallels, a history of brain development, thought experiments, and much more.

The article tells how such a system can bring the thinking and the system of knowledge of robots closer to the human. And what dreams do robots see today?

As with the help of the vector can indicate the semantic meaning and which models should be used in this case. The process of learning the network, open source and description of the result - the recipe for a good article.

An ITMO University article on how the use of neural networks improves the effectiveness of an attack detection system and increases the speed at which they are detected. Two approaches are considered: sharing NA with expert systems and using NA as a separate system.

Medicine, financial services, education, production and the film industry: what problems does AI solve, who is behind the revolutionary solutions in each industry and what to expect in the future?

Artificial intelligence greatly simplifies the design process and minimizes decision-making time at each stage. In addition to the detailed theoretical part, the article presents an example of the practical implementation of the project.