Dominion: A New Frontier for AI Research

Read original: arXiv:2405.06846 - Published 5/14/2024 by Danny Halawi, Aron Sarmasi, Siena Saltzen, Joshua McCoy
Total Score

0

Dominion: A New Frontier for AI Research

Sign in to get full access

or

If you already have an account, we'll log you in

Overview

  • This paper introduces Dominion, a new environment for AI research that offers a rich and complex game-playing domain.
  • Dominion is a deck-building card game that combines elements of strategy, resource management, and randomness, presenting unique challenges for AI systems.
  • The paper describes the Dominion Online Dataset, a large-scale dataset of human gameplay data that can be used to train and evaluate AI models in this domain.

Plain English Explanation

The paper presents a new frontier for AI research called Dominion, which is a deck-building card game. Dominion is a complex game that requires players to strategically manage resources and adapt to randomness, making it an interesting challenge for AI systems. The researchers have created a large dataset of human gameplay data from the online version of Dominion, which can be used to train and test AI models on this domain.

Deck-building games like Dominion: A New Frontier for AI Research combine elements of strategy, resource management, and randomness, providing a rich environment for AI research. The Dominion Online Dataset can help researchers develop AI systems that can excel at complex game-playing tasks, similar to how CNN-based approaches have been used to tackle other game-related challenges. Overall, this research opens up new avenues for exploring zero-sum positional differential games and advancing the field of AI.

Technical Explanation

The paper introduces Dominion, a deck-building card game, as a new environment for AI research. Dominion combines elements of strategy, resource management, and randomness, presenting unique challenges for AI systems. The researchers have created the Dominion Online Dataset, a large-scale dataset of human gameplay data from the online version of the game, which can be used to train and evaluate AI models in this domain.

The Dominion game involves players building their own decks of cards, with each card providing different abilities and resources. Players must strategically manage their deck, purchase new cards, and adapt to the randomness of card draws and other players' actions. This game-playing environment offers a rich and complex challenge for AI systems, going beyond traditional board games or video games.

The Dominion Online Dataset contains a vast amount of gameplay data, including player actions, card interactions, and game outcomes. This dataset can be used to train and evaluate AI models that can learn to play Dominion at a high level, mimicking and potentially surpassing human performance. The dataset's size and diversity can help researchers develop robust and adaptable AI systems that can excel in this complex game-playing domain.

Critical Analysis

The paper presents a promising new frontier for AI research, but there are a few potential limitations and areas for further exploration:

  1. The complexity of Dominion may present significant challenges for current AI techniques, especially in terms of handling the game's randomness and the vast decision space. The researchers may need to explore more advanced reinforcement learning or hybrid approaches to tackle these challenges effectively.

  2. The Dominion Online Dataset, while extensive, may not capture the full range of human strategies and decision-making processes. It would be valuable to supplement this dataset with additional human gameplay data, potentially from expert players or tournaments, to ensure the AI models can learn from a diverse set of gameplay behaviors.

  3. The paper does not discuss the potential societal implications of developing highly capable Dominion-playing AI systems. As these systems become more advanced, it will be important to consider their potential impact on game-playing communities, as well as any broader ethical considerations.

Overall, the introduction of Dominion as a new AI research environment is an exciting development, and the Dominion Online Dataset provides a valuable resource for the research community. However, the complexity of the game and the need for a more comprehensive understanding of human gameplay behaviors will require continued research and innovation to fully realize the potential of this new frontier.

Conclusion

This paper presents Dominion, a deck-building card game, as a new and promising environment for AI research. Dominion's combination of strategy, resource management, and randomness offers unique challenges for AI systems, going beyond traditional game-playing domains. The creation of the Dominion Online Dataset, a large-scale dataset of human gameplay data, provides a valuable resource for training and evaluating AI models in this complex game-playing domain.

The potential of Dominion as an AI research environment is significant, as it can push the boundaries of current AI techniques and lead to the development of more robust, adaptable, and high-performing game-playing systems. Additionally, the insights gained from this research could have broader implications for the field of AI, contributing to the advancement of reinforcement learning, decision-making, and reasoning capabilities. As the research in this domain continues, it will be important to address the challenges posed by the game's complexity and to consider the potential societal implications of these AI systems.



This summary was produced with help from an AI and may contain inaccuracies - check out the links to read the original source documents!

Follow @aimodelsfyi on 𝕏 →

Related Papers

Dominion: A New Frontier for AI Research
Total Score

0

Dominion: A New Frontier for AI Research

Danny Halawi, Aron Sarmasi, Siena Saltzen, Joshua McCoy

In recent years, machine learning approaches have made dramatic advances, reaching superhuman performance in Go, Atari, and poker variants. These games, and others before them, have served not only as a testbed but have also helped to push the boundaries of AI research. Continuing this tradition, we examine the tabletop game Dominion and discuss the properties that make it well-suited to serve as a benchmark for the next generation of reinforcement learning (RL) algorithms. We also present the Dominion Online Dataset, a collection of over 2,000,000 games of Dominion played by experienced players on the Dominion Online webserver. Finally, we introduce an RL baseline bot that uses existing techniques to beat common heuristic-based bots, and shows competitive performance against the previously strongest bot, Provincial.

Read more

5/14/2024

AlphaDou: High-Performance End-to-End Doudizhu AI Integrating Bidding
Total Score

0

AlphaDou: High-Performance End-to-End Doudizhu AI Integrating Bidding

Chang Lei, Huan Lei

Artificial intelligence for card games has long been a popular topic in AI research. In recent years, complex card games like Mahjong and Texas Hold'em have been solved, with corresponding AI programs reaching the level of human experts. However, the game of Doudizhu presents significant challenges due to its vast state/action space and unique characteristics involving reasoning about competition and cooperation, making the game extremely difficult to solve.The RL model Douzero, trained using the Deep Monte Carlo algorithm framework, has shown excellent performance in Doudizhu. However, there are differences between its simplified game environment and the actual Doudizhu environment, and its performance is still a considerable distance from that of human experts. This paper modifies the Deep Monte Carlo algorithm framework by using reinforcement learning to obtain a neural network that simultaneously estimates win rates and expectations. The action space is pruned using expectations, and strategies are generated based on win rates. The modified algorithm enables the AI to perform the full range of tasks in the Doudizhu game, including bidding and cardplay. The model was trained in a actual Doudizhu environment and achieved state-of-the-art performance among publicly available models. We hope that this new framework will provide valuable insights for AI development in other bidding-based games.

Read more

9/16/2024

🤿

Total Score

0

Deep Reinforcement Learning for 5*5 Multiplayer Go

Brahim Driss, J'er^ome Arjonilla, Hui Wang, Abdallah Saffidine, Tristan Cazenave

In recent years, much progress has been made in computer Go and most of the results have been obtained thanks to search algorithms (Monte Carlo Tree Search) and Deep Reinforcement Learning (DRL). In this paper, we propose to use and analyze the latest algorithms that use search and DRL (AlphaZero and Descent algorithms) to automatically learn to play an extended version of the game of Go with more than two players. We show that using search and DRL we were able to improve the level of play, even though there are more than two players.

Read more

5/24/2024

Games of Knightian Uncertainty
Total Score

0

Games of Knightian Uncertainty

Spyridon Samothrakis, Dennis J. N. J. Soemers, Damian Machlanski

Arguably, for the latter part of the late 20th and early 21st centuries, games have been seen as the drosophila of AI. Games are a set of exciting testbeds, whose solutions (in terms of identifying optimal players) would lead to machines that would possess some form of general intelligence, or at the very least help us gain insights toward building intelligent machines. Following impressive successes in traditional board games like Go, Chess, and Poker, but also video games like the Atari 2600 collection, it is clear that this is not the case. Games have been attacked successfully, but we are nowhere near AGI developments (or, as harsher critics might say, useful AI developments!). In this short vision paper, we argue that for game research to become again relevant to the AGI pathway, we need to be able to address textit{Knightian uncertainty} in the context of games, i.e. agents need to be able to adapt to rapid changes in game rules on the fly with no warning, no previous data, and no model access.

Read more

6/28/2024