Pub Date : 2015-08-01DOI: 10.1109/CIG.2015.7317659
Martin Müller
With the success of Monte Carlo Tree Search, the game of Go has become a focus of games research. Recently, deep convolutional neural networks have achieved human-level performance in predicting master moves. Even before that, machine learning techniques have been used very successfully as an automated way to improve the domain knowledge in Go programs. Go programs have now reached a level close to top amateur players. In order to challenge professional level players, we must combine the three pillars of modern Go programs — search, knowledge, and simulation — in a high performance system, possibly running on massively parallel hardware. This talk will summarize recent progress in this exciting field, and outline a research strategy for boosting the performance of Go programs to the next level.
{"title":"Keynote speech III: Computer go research - The challenges ahead","authors":"Martin Müller","doi":"10.1109/CIG.2015.7317659","DOIUrl":"https://doi.org/10.1109/CIG.2015.7317659","url":null,"abstract":"With the success of Monte Carlo Tree Search, the game of Go has become a focus of games research. Recently, deep convolutional neural networks have achieved human-level performance in predicting master moves. Even before that, machine learning techniques have been used very successfully as an automated way to improve the domain knowledge in Go programs. Go programs have now reached a level close to top amateur players. In order to challenge professional level players, we must combine the three pillars of modern Go programs — search, knowledge, and simulation — in a high performance system, possibly running on massively parallel hardware. This talk will summarize recent progress in this exciting field, and outline a research strategy for boosting the performance of Go programs to the next level.","PeriodicalId":6594,"journal":{"name":"2016 IEEE Conference on Computational Intelligence and Games (CIG)","volume":"26 1","pages":"18"},"PeriodicalIF":0.0,"publicationDate":"2015-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83087583","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2015-08-01DOI: 10.1109/CIG.2015.7317658
S. Lucas
Although AI has excelled at many narrowly defined problems, it is still very far from achieving human-like performance in terms of solving problems that it was not specifically programmed for: hence the challenge of artificial general intelligence (AGI) was developed to foster more general AI research. A promising way to address this is to pose the challenge of learning to play video games without knowing any details of the games in advance. In order to study this in a systematic way the General Video Game AI (http://gvgai.net) competition series was created. This provides an excellent challenge for computational intelligence and AI methods and initial results indicate often good though somewhat patchy performance from simulation-based methods such as Monte Carlo Tree Search and Rolling Horizon Evolutionary Algorithms. Observing where these methods succeed and fail leads to the conclusion that there is still much scope for further developing algorithms that mix simulation with long-term learning. While running the competitions we have built up a large set of GVGAI players. This large pool of adaptive players leads on to very appealing potential applications in automated and semi-automated game design where the player-set can be used to evaluate novel games and new parameter settings of existing games. Initial explorations of this idea will be discussed.
尽管人工智能在许多狭义的问题上表现出色，但在解决没有专门为其编程的问题方面，它还远未达到类似人类的表现:因此，开发通用人工智能(AGI)的挑战是为了促进更通用的人工智能研究。解决这一问题的一个有效方法是，让玩家在事先不了解任何游戏细节的情况下学习玩电子游戏。为了以一种系统的方式研究这一点，我们创造了General Video Game AI (http://gvgai.net)竞赛系列。这为计算智能和人工智能方法提供了一个极好的挑战，最初的结果表明，基于模拟的方法(如蒙特卡罗树搜索和滚动地平线进化算法)的性能通常很好，尽管有些不完整。通过观察这些方法的成功和失败，我们可以得出结论:将模拟与长期学习相结合的算法仍有很大的发展空间。在举办比赛的过程中，我们已经建立了一大批GVGAI玩家。这一大群自适应玩家将在自动化和半自动化游戏设计中带来非常有吸引力的潜在应用，即玩家集可用于评估新游戏和现有游戏的新参数设置。我们将讨论对这一想法的初步探索。
{"title":"Keynote speech II: General video game AI: Challenges and applications","authors":"S. Lucas","doi":"10.1109/CIG.2015.7317658","DOIUrl":"https://doi.org/10.1109/CIG.2015.7317658","url":null,"abstract":"Although AI has excelled at many narrowly defined problems, it is still very far from achieving human-like performance in terms of solving problems that it was not specifically programmed for: hence the challenge of artificial general intelligence (AGI) was developed to foster more general AI research. A promising way to address this is to pose the challenge of learning to play video games without knowing any details of the games in advance. In order to study this in a systematic way the General Video Game AI (http://gvgai.net) competition series was created. This provides an excellent challenge for computational intelligence and AI methods and initial results indicate often good though somewhat patchy performance from simulation-based methods such as Monte Carlo Tree Search and Rolling Horizon Evolutionary Algorithms. Observing where these methods succeed and fail leads to the conclusion that there is still much scope for further developing algorithms that mix simulation with long-term learning. While running the competitions we have built up a large set of GVGAI players. This large pool of adaptive players leads on to very appealing potential applications in automated and semi-automated game design where the player-set can be used to evaluate novel games and new parameter settings of existing games. Initial explorations of this idea will be discussed.","PeriodicalId":6594,"journal":{"name":"2016 IEEE Conference on Computational Intelligence and Games (CIG)","volume":"38 1","pages":"17"},"PeriodicalIF":0.0,"publicationDate":"2015-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76931449","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2014-01-01DOI: 10.1109/CIG.2014.6932871
Paolo Burelli, G. Triantafyllidis, I. Patras
{"title":"Proceedings of the 2014 IEEE Conference on Computational Intelligence and Games","authors":"Paolo Burelli, G. Triantafyllidis, I. Patras","doi":"10.1109/CIG.2014.6932871","DOIUrl":"https://doi.org/10.1109/CIG.2014.6932871","url":null,"abstract":"","PeriodicalId":6594,"journal":{"name":"2016 IEEE Conference on Computational Intelligence and Games (CIG)","volume":"1 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2014-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86280329","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2011-09-29DOI: 10.1109/CIG.2011.6031995
Jacob Schrum, R. Miikkulainen
Intelligent opponent behavior helps make video games interesting to human players. Evolutionary computation can discover such behavior, especially when the game consists of a single task. However, multitask domains, in which separate tasks within the domain each have their own dynamics and objectives, can be challenging for evolution. This paper proposes two methods for meeting this challenge by evolving neural networks: 1) Multitask Learning provides a network with distinct outputs per task, thus evolving a separate policy for each task, and 2) Mode Mutation provides a means to evolve new output modes, as well as a way to select which mode to use at each moment. Multitask Learning assumes agents know which task they are currently facing; if such information is available and accurate, this approach works very well, as demonstrated in the Front/Back Ramming game of this paper. In contrast, Mode Mutation discovers an appropriate task division on its own, which may in some cases be even more powerful than a human-specified task division, as shown in the Predator/Prey game of this paper. These results demonstrate the importance of both Multitask Learning and Mode Mutation for learning intelligent behavior in complex games.
{"title":"Evolving multimodal networks for multitask games","authors":"Jacob Schrum, R. Miikkulainen","doi":"10.1109/CIG.2011.6031995","DOIUrl":"https://doi.org/10.1109/CIG.2011.6031995","url":null,"abstract":"Intelligent opponent behavior helps make video games interesting to human players. Evolutionary computation can discover such behavior, especially when the game consists of a single task. However, multitask domains, in which separate tasks within the domain each have their own dynamics and objectives, can be challenging for evolution. This paper proposes two methods for meeting this challenge by evolving neural networks: 1) Multitask Learning provides a network with distinct outputs per task, thus evolving a separate policy for each task, and 2) Mode Mutation provides a means to evolve new output modes, as well as a way to select which mode to use at each moment. Multitask Learning assumes agents know which task they are currently facing; if such information is available and accurate, this approach works very well, as demonstrated in the Front/Back Ramming game of this paper. In contrast, Mode Mutation discovers an appropriate task division on its own, which may in some cases be even more powerful than a human-specified task division, as shown in the Predator/Prey game of this paper. These results demonstrate the importance of both Multitask Learning and Mode Mutation for learning intelligent behavior in complex games.","PeriodicalId":6594,"journal":{"name":"2016 IEEE Conference on Computational Intelligence and Games (CIG)","volume":"8 1","pages":"102-109"},"PeriodicalIF":0.0,"publicationDate":"2011-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83577171","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2010-09-30DOI: 10.1109/ITW.2010.5593384
Johan Hagelbäck, S. Johansson, M. Preuss
Real-time strategy games (RTS) are an active area of research as well as a popular branch of industrial game production, with high commercial interest. Although player satisfaction is the ultimate goal also for these games, they are usually too complex to come up with human-level AI that is not cheating. In consequence, for RTS games it is as desirable to play well as it is to make the game interesting. Also, RTS games have many aspects that call for CI or other innovative methods, as strategy, tactics, resource management, and many more. The task of the special session is to advance the state of research on RTS by new methods or new applications of methods, new concepts, and also the analysis of existing methods or problems.
{"title":"AI and computational intelligence for real-time strategy games","authors":"Johan Hagelbäck, S. Johansson, M. Preuss","doi":"10.1109/ITW.2010.5593384","DOIUrl":"https://doi.org/10.1109/ITW.2010.5593384","url":null,"abstract":"Real-time strategy games (RTS) are an active area of research as well as a popular branch of industrial game production, with high commercial interest. Although player satisfaction is the ultimate goal also for these games, they are usually too complex to come up with human-level AI that is not cheating. In consequence, for RTS games it is as desirable to play well as it is to make the game interesting. Also, RTS games have many aspects that call for CI or other innovative methods, as strategy, tactics, resource management, and many more. The task of the special session is to advance the state of research on RTS by new methods or new applications of methods, new concepts, and also the analysis of existing methods or problems.","PeriodicalId":6594,"journal":{"name":"2016 IEEE Conference on Computational Intelligence and Games (CIG)","volume":"7 1","pages":"1-4"},"PeriodicalIF":0.0,"publicationDate":"2010-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87383810","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2010-01-01DOI: 10.1109/ITW.2010.5593319
Emilio Martín, Moisés Martínez, Gustavo Recio, Y. Sáez
{"title":"Pac-mAnt: Optimization based on ant colonies applied to developing an agent for Ms. Pac-Man","authors":"Emilio Martín, Moisés Martínez, Gustavo Recio, Y. Sáez","doi":"10.1109/ITW.2010.5593319","DOIUrl":"https://doi.org/10.1109/ITW.2010.5593319","url":null,"abstract":"","PeriodicalId":6594,"journal":{"name":"2016 IEEE Conference on Computational Intelligence and Games (CIG)","volume":"75 1","pages":"458-464"},"PeriodicalIF":0.0,"publicationDate":"2010-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79601392","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Traditionally, the game of Tic-tac-toe is a pencil and paper game played by two people who take turn to place their pieces on a 3times3 grid with the objective of being the first player to fill a horizontal, vertical, or diagonal row with their pieces. What if instead of having one person playing against another, one person plays against a team of nine players, each of whom is responsible for one cell in the 3times3 grid? In this new way of playing the game, the team has to coordinate its players, who are acting independently based on their limited information. In this paper, we present a solution that can be extended to the case where two such teams play against each other, and also to other board games. Essentially, the solution uses a decentralized decision making, which at first seems to complicate the solution. However, surprisingly, we show that in this mode, an equivalent level of decision making ability comes from simple components that reduce system complexity
{"title":"Decentralized Decision Making in the Game of Tic-tac-toe","authors":"E. Soedarmadji","doi":"10.1109/CIG.2006.311678","DOIUrl":"https://doi.org/10.1109/CIG.2006.311678","url":null,"abstract":"Traditionally, the game of Tic-tac-toe is a pencil and paper game played by two people who take turn to place their pieces on a 3times3 grid with the objective of being the first player to fill a horizontal, vertical, or diagonal row with their pieces. What if instead of having one person playing against another, one person plays against a team of nine players, each of whom is responsible for one cell in the 3times3 grid? In this new way of playing the game, the team has to coordinate its players, who are acting independently based on their limited information. In this paper, we present a solution that can be extended to the case where two such teams play against each other, and also to other board games. Essentially, the solution uses a decentralized decision making, which at first seems to complicate the solution. However, surprisingly, we show that in this mode, an equivalent level of decision making ability comes from simple components that reduce system complexity","PeriodicalId":6594,"journal":{"name":"2016 IEEE Conference on Computational Intelligence and Games (CIG)","volume":"107 1","pages":"34-38"},"PeriodicalIF":0.0,"publicationDate":"2006-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86245129","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：