The best footballers are not necessarily the ones with the best physical skills. The difference between success and failure in football often lies in the ability to make the right decisions on the field in a split second regarding the race location and the appropriate time to attack, pass or otherwise. shoot.
How can clubs help players train their brains and their bodies?
We do this by analyzing several seasons of data that follow the players and the ball at each game and developing a computer model of different game positions.
The computer model provides a reference for comparing the performance of different actors. In this way, we can measure the performance of individual players regardless of the actions of other players.
We can then see what could have happened if the players had made a different decision in any case. TV commentators still criticize players' actions, saying that they should have done something else without actually testing the theory. But our computer model can show how realistic these suggestions might be.
If a reviewer indicates that a player should have dribbled instead of passing, our system may examine the alternative result, taking into account factors such as the state of fatigue of the player at this stage of the game .
We hope that coaches and support staff will use the system to help players reflect on their actions after a game and, over time, improve their decision-making skills.
Modeling of decision-making
Measuring these skills is extremely difficult for a number of reasons. First, a human can not follow all the events that take place during a match. Secondly, it is difficult to isolate a player's actions from those of another.
For example, if a player passes the ball and a few seconds later, the team loses the ball, was the player mistimed to the wrong person or was it someone's fault? another one?
To solve this problem, we use a specific branch of the AI known as imitation learning . This technology makes it possible to learn computer models of behavior, such as the actions of footballers in the field, by analyzing huge amounts of historical data.
In simple terms, the computer model learns to imitate human experts.
Most artificial intelligence decision systems, such as those used to play board games as Go are based on reinforcement learning. It's here that a computer learns to make decisions by trying several times until it is informed that it has acted well, much as if one were training a dog to do something by rewarding him.
But most real world scenarios do not offer a specific reward like winning in a board game.
Learning by imitation, by contrast, attempts to understand the underlying decision-making policy. by looking at how an expert performs a task, then tries to imitate it.
It is very difficult to model football experts (players) because they make decisions with advanced skills that are difficult to program in a computer, such as choosing the items to pay attention to, selecting the correct answer, and anticipating what other players will do. make.
So for the computer model to be realistic, the historical data on which it is based must reflect the real world as much as possible. This should not simply show how players move relative to each other and the ball, but also reflect their fatigue and play situation.
For example, do players want to attack or attempt to defend themselves, or even if they want to win or lose. (In some tournaments, a team might want to lose a match so that its position in the next round gives it an easier opponent.)
Modification of post match analysis
We have already developed a system that can create a pattern of player movement relative to each other and a balloon that can be used to study performance.
We now plan to make the model more realistic by adding details of the players' body posture, their heart rate (to represent fatigue) and the playing conditions. We will then develop the system to measure the skills of the players. current players and hope to have a fully functional system within two years.
We expect players and coaches to analyze their games, especially post game scans. This will help players to be more reflective by being able to see how their actions could have made a difference. Scouts and clubs would be able to select players and identify talents using data relating to these critical decision-making skills.
The extension of AI from controlled gaming environments to complex real-world applications remains a daunting challenge. But humans are very good at adapting and making decisions in complex and changing environments.
Thus, by learning to emulate human decision-making, AI will be able to deal with all kinds of unfamiliar environments in which people do not always follow the rules.
This article is republished from The Conversation of Varuna De Silva Lecturer, Institute for Digital Technologies, of Loughborough University under a Creative Commons license. Read the original article .