Talks
Speakers
Events
Topics
Sign in
Home
Talks
Speakers
Events
Topics
Leaderboard
Use
Analytics
Sign in
Suggest modification to this talk
Title
Description
Beating Go thanks to the power of randomness by Tobias Pfeiffer Go is a board game that is more than 2,500 years old (yes, this is not about the programming language!) and it is fascinating from multiple viewpoints. For instance, go bots still can’t beat professional players, unlike in chess. This talk will show you what is so special about Go that computers still can’t beat humans. We will take a look at the most popular underlying algorithm and show you how the Monte Carlo method, basically random simulation, plays a vital role in conquering Go's complexity and creating the strong Go bots of today. Help us caption & translate this video! http://amara.org/v/H1VW/
Date
Summarized using AI?
If this talk's summary was generated by AI, please check this box. A "Summarized using AI" badge will be displayed in the summary tab to indicate that the summary was generated using AI.
Show "Summarized using AI" badge on summary page
Summary
Markdown supported
In the talk "Beating Go Thanks To The Power Of Randomness," presented by Tobias Pfeiffer at RubyConf 2015, the speaker explores the fascinating complexity of the board game Go, which has resisted AI dominance unlike chess. The key topics covered in the presentation include: - **Historical Context**: The speaker discusses pivotal moments in AI history, including IBM's Deep Blue defeating Garry Kasparov in chess in 1997, and the events surrounding Go, such as programs playing against strong young players and the development of significant Go programs. - **Complexity of Go**: Go’s rules are simpler than chess, but the game’s strategic depth and the vast number of potential board configurations (estimated to be much higher than chess) make it notably more complex. Key statistics highlight that Go has a branching factor of around 250, compared to chess's 35. - **Game Mechanics**: The speaker explains the basics of Go, including the concept of liberties, capture, and territory control, as well as the importance of komi used to balance the first-move advantage. - **Differences from Chess**: In chess, players focus on dismantling their opponent’s setup, whereas in Go, every move contributes to building the board. This leads to a more dynamic playing experience, where moves have global impacts. - **AI Challenges**: Traditional AI algorithms like Min-Max used in chess struggle with Go’s complexity, prompting the exploration of the Monte Carlo method, which utilizes randomness to estimate outcomes. - **Monte Carlo Method**: The speaker highlights this approach, explaining its phases such as selection, expansion, simulation, and backpropagation. An analogy involving casino games illustrates the balance between exploring new possibilities and exploiting known strategies. - **Neural Networks & AI Development**: The incorporation of neural networks into the Monte Carlo framework is discussed as a promising avenue for improving AI performance in Go, providing a more nuanced approach to predicting strategy. - **Engagement with Technology**: The talk emphasizes the excitement surrounding the shared exploration of gaming, particularly in terms of developing engines and enhancing strategies. In conclusion, Tobias Pfeiffer’s talk sheds light on the intricacies of Go, the limitations faced by AI, and how randomness can be harnessed to improve strategy development, illustrating the continued evolution and interest in this ancient game.
Suggest modifications
Cancel
