In the 21st century, game theory explains many kinds of behaviour. It is now an umbrella term for the science of how humans, animals, and computers make logical decisions.
Game theory studies mathematical models of how rational people act in a group. We can use it in all areas of social science, as well as in logic, systems science, and computer science.
Take the support vector machine (SVM) as an example. Game theory says that the SVM is a two-player game in which one player challenges the other to find the best hyper-plane after giving the most complex points to classify. The last round of this game will come down to a trade-off between the strategic skills of the two players.
Game theory can be in different areas of artificial intelligence:
- Multi-agent AI systems.
- Imitation and Reinforcement Learning.
- Adversary training in Generative Adversarial Networks (GANs).
History
During the 1950s, many scholars worked on game theory a lot. The researchers first used game theory to talk about evolution in the 1970s. But similar ideas have been around since at least the 1930s.
Game theory can also explain many things happening to us daily and how Machine Learning works. For example, a classification algorithm like SVM (Support Vector Machines) can be a two-player game in which one player challenges the other to find the best hyperplane by giving him the most difficult points to classify. The game will then settle on a solution that is a trade-off between the two players’ strategic skills.
Furthermore, researchers made most of the popular digital games we play with the help of AI and game theory. However, game theory isn’t just limited to games; it also has many AI applications, such as
- Generative Adversarial Networks (GANs),
- machine-learning algorithms,
- manipulation-resistant systems,
- multi-agent AI systems, imitation and
- reinforcement learning, etc.
Types of Games theory
The game theory covers various games, depending on the players and what they do. Single-move Games are the most important and well-known ones, like buying stocks. Repeated Games, like the Prisoner’s Dilemma, and Sequential Games, like Chess and Go.
There are approximately five types of game classification at present. They are listed below:
- Zero-Sum and Non-Zero Sum Games,
- Simultaneous and Sequential Games,
- Imperfect Information and Perfect Information Games,
- Asymmetric and Symmetric Games,
- Co-operative and Non Co-operative Games
Inverse Game Theory
Game theory aims to comprehend a game’s dynamics to maximize the feasible outcome for its players. Instead, Inverse Game Theory seeks to build a game based on the strategies and objectives of the players. Therefore, in developing habitats for AI Agents, Inverse Game Theory plays a crucial role.
Conclusion
In multi-agent reinforcement learning, AI-controlled self-driving cars are to improve the flow of traffic in an area with the help of game theory.
Moreover, game theory is also different from machine learning, which has real-world and everyday effects. So we can guess the basic idea of game theory from many things that happen in real life.
Source: indiaai.gov.in
