Tips for Lippi: National Football goes further, try artificial intelligence

A game with multiple meanings came to an end last night. The national team played in Changsha and beat South Korea.

“The sky is bright”, said Li Yi, the great friend in the circle of friends and Weibo. The circle of friends and Weibo have also been brushed by the national football team.

Lippi said after the game: "I hope my team will not only meet today's performance, hoping that the latter game will perform better than today." How can we be more excellent?

"If you don't know enough, then you can also rebel; How to examine their own deficiencies, more intuitive to see the gap with their opponents? And how to pick the best player in the state?

With all of these issues, artificial intelligence can help give answers.

Insufficient knowledge by deep imitation learning

Take Fulham's Premier League match against Swansea as an example. Fulham (red) is attacking and Swansea (blue) is defending. In the actual game, Fulham scored the goal into Swansea's goal.

We analyze from a defensive point of view: How much room does Swansea still have to improve?

Take a look at the answer to artificial intelligence. This artificial intelligence system is based on deep imitation learning. The principle is similar to DeepMind's AlphaGo. It is also a neural network based on data training.

The left side of the figure above is the actual game situation. The model predicts that Swansea's probability of losing the ball is 69.1%. The middle part is the defensive position (white) simulated by the AI ​​according to the average level of the Premier League. The model predicts that the probability of losing the ball is 71.8%. The right is based on Manchester City's data, AI mimics the defensive position, the probability of conceding to 41.7%.

This process can be more clearly demonstrated with the following diagram:

This neural network uses 100 games in the Premier League for training. Each player is modeled by LSTM and includes two hidden layers, each with 512 neurons. This algorithm also simulates the interaction between teammates.

With more and more data accumulated on the field, with the help of AI, the head coach can find the team's weakness more precisely and more clearly, and he can get the analysis result without investing a lot of manpower.

1 2 3 Next>

Fiber Optic Splice Box

Fibre Optic Cable Joint Box,Outdoor Fiber Splice Box,Optical Fiber Cable Joint Box,Fusion Splice Box

Ningbo Fengwei Communication Technology Co., Ltd , https://www.fengweicommunication.com