Liverpool FC and DeepMind are exploring the use of neural networks in football
It was revealed that Liverpool Football Club has teamed up with the development team of Alphabet-owned DeepMind to explore the possibilities of using artificial intelligence in the world of football. The results of the joint work were published in the Journal of Artificial Intelligence Research.
In recent years, the amount of data available in football has increased significantly with various sensors, GPS trackers, and computer vision algorithms that are used to track the movement of the ball and the movement of players across the field. The abundance of data means that advanced technologies can be used to analyze it.
Research has shown that football teams can use neural networks to identify patterns that a coach cannot see. It is noted that football offers DeepMind a limited but challenging environment for testing algorithms. “A game like football is exciting because there are many agents, there are aspects of competition and cooperation. Unlike chess or go, football is inherently uncertain because it is played in an open world, ”said Karl Tuyls, one of the researchers at DeepMind.
Uncertainty doesn’t mean you can’t make predictions in football, and this is one area where AI can be most useful. For example, you can train a neural network using data about a specific team and line-up to predict how players will react in different situations on the field. Such an algorithm will predict the consequences of tactical changes or how, for example, the opponent will play if a key player leaves the field due to injury.
The purpose of the study is not to create a replacement for coaches but to create an additional tool. “There is a large amount of data that the algorithm has to digest, and it is not always easy to process such data arrays. We are trying to create assistive technology, ”the researchers say. The neural network could also be used to predict how a particular player will take a penalty. Another option for using neural networks in football is to assess the impact of specific actions of players in a given situation on the final result. Such information would be useful for post-match analysis.
“The goal is to have a holistic system that integrates well with the human player on the pitch and makes his job easier. I do not think that you will see great progress in the next six months or a year, but some tools will be more developed in the next five years. You will see the emergence of something like “Automated videopomoschnika for a coach” who will carry out pre-match and Post-match analysis “– considers Karl Styles.
