Artificial Intelligence and Esports
When Machine Learning ‘learns’ through play
Reading Time: 3 minutes
Post published on 11/11/2020 by Donata Petrelli and released with licenza CC BY-NC-ND 3.0 IT (Creative Common – Attribuzione – Non commerciale – Non opere derivate 3.0 Italia)
Title Image credits by Jakob Wells on Flickr
Who said that Artificial Intelligence isn’t fun?
With Esports, the digital version of competitive sports, Artificial Intelligence “starts playing”, thus opening up technology to completely new and unexplored scenarios.
In this short article we will see how Sport is a field of application of state-of-the-art Machine Learning technologies and how that could be emulated by other sectors.
From talent scouts to statistical data analysis systems
The figure of the talent scout has always assumed strategic importance in many professional clubs. For example, in America it is a recognised profession in the NBA!
This is because the first and most important analysis comes from the analytical observation of a phenomenon. So observing matches played by youth and budding teams allows you to find players with great potential and future champions to spend in your team.
Now analytical observation alone is no longer enough and we have moved on to statistical data analysis. The study of players no longer takes place on the playing field but on a database. The film “The Art of Winning” (or the book Moneyball) tells how the general manager of the Oakland Athletics decides to adopt an alternative system on how to evaluate a player based on analysis through statistics. Through this system it is intended to establish a player’s value from performance in past seasons to try to predict his future value. This innovative method has proven to be effective in measuring the sporting performance of players.
The same thing happens in many sectors, including the financial markets for trend analysis. It starts with the analysis of historical data to arrive, through pre-processing techniques and data mining, to study the underlying trend and to estimate the probability of future values.
Statistical techniques can be useful whether it is a question of prices or volumes of stock market assets, or of shots on goal or matches won.
With the great diffusion of Artificial Intelligence, many disciplines have been converted to it. Sport was no exception!
The evaluation of the historical series is thus entrusted to Artificial Intelligence. The result of the operation can be a ranking. Depending on the analysis parameters chosen, values are obtained for the players, according to their role or field area, or the technical growth trend can be assessed to estimate their future potential.
There are many similarities with Sport. For example, in the field of Robo Advisoring, IA is used to create investment portfolios that meet the user’s needs in terms of time and risk. And so, at the rankings of players drawn up for objectives in parallel we have portfolios, rankings of financial instruments that are designed to achieve certain returns.
With time technology advances and the frontiers of its applications still open up. Data and Big Data seems to be a thing of the past… one prefers to move from brute data analysis to experience. And so we are at the Machine Learning stage, the Artificial Intelligence that learns from experience and its own mistakes to obtain realistic models from which to draw all the necessary information. It’s the age of the Esports!
Esports are the digital version of competitive sports, today’s biggest entertainment and gaming phenomenon. It is interesting for us, however, the use that is made of them, that is to make Artificial Intelligence experience a world that, although digital, is realistic.
We are not talking about robots in the real world, endowed with sense organs to interact with them, but about intelligent software that plays in a real (but virtual) world and is suitable for learning.
It is significant that it is precisely the game that is the first meeting ground between artificial and real intelligences.
Suffice it to say that OpenAI, the non-profit organization for research on artificial intelligence in San Francisco, has invested a lot in this field creating software able to play according to the same rules as human players. In contrast to games like chess based on speed analysis, Machine Learning techniques are used here.
Artificial Intelligence plays against itself by learning from itself game techniques identifiable to those of human players.
In addition to games … in what other fields of application could we take advantage of the logic of Esports?