Pub Date : 2016-09-01DOI: 10.1109/CIG.2016.7860391
Daniel Karavolos, Antonios Liapis, Georgios N. Yannakakis
This paper describes a search-based level generation approach that uses the search space of action sequences, represented as graphs, rather than spatial layouts. The search is guided by mutation operators that manipulate the graph topology, and the paper explores various objective functions that are based on generic level evaluation metrics. The evolved action sequences are passed to a grammar-based system and a layout solver transforms them into dungeon levels for the Dwarf Quest game.
{"title":"Evolving missions for Dwarf quest dungeons","authors":"Daniel Karavolos, Antonios Liapis, Georgios N. Yannakakis","doi":"10.1109/CIG.2016.7860391","DOIUrl":"https://doi.org/10.1109/CIG.2016.7860391","url":null,"abstract":"This paper describes a search-based level generation approach that uses the search space of action sequences, represented as graphs, rather than spatial layouts. The search is guided by mutation operators that manipulate the graph topology, and the paper explores various objective functions that are based on generic level evaluation metrics. The evolved action sequences are passed to a grammar-based system and a layout solver transforms them into dungeon levels for the Dwarf Quest game.","PeriodicalId":6594,"journal":{"name":"2016 IEEE Conference on Computational Intelligence and Games (CIG)","volume":"68 1","pages":"1-2"},"PeriodicalIF":0.0,"publicationDate":"2016-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82552585","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 : 2016-09-01DOI: 10.1109/CIG.2016.7860429
C. F. Sironi, M. Winands
General Game Playing (GGP) aims at creating computer programs able to play any arbitrary game at an expert level given only its rules. The lack of game-specific knowledge and the necessity of learning a strategy online have made Monte-Carlo Tree Search (MCTS) a suitable method to tackle the challenges of GGP. An efficient search-control mechanism can substantially increase the performance of MCTS. The RAVE strategy and its more recent variant, GRAVE, have been proposed for this reason. In this paper we further investigate the use of GRAVE for GGP and compare its performance with the more established RAVE strategy and with a new variant, called HRAVE, that uses more global information. Experiments show that for some games GRAVE and HRAVE perform better than RAVE, with GRAVE being the most promising one overall.
通用游戏玩法(General Game Playing，简称GGP)的目标是创建能够在给定规则的情况下以专家水平玩任意游戏的计算机程序。缺乏游戏特定知识和在线学习策略的必要性使得蒙特卡洛树搜索(MCTS)成为解决GGP挑战的合适方法。有效的搜索控制机制可以大大提高MCTS的性能。RAVE策略及其最近的变体GRAVE正是出于这个原因而被提出的。在本文中，我们进一步研究了GRAVE在GGP中的使用，并将其性能与更成熟的RAVE策略以及使用更多全局信息的新变体HRAVE进行了比较。实验表明，在某些游戏中，GRAVE和HRAVE的表现优于RAVE，其中GRAVE是最有前途的一个。
{"title":"Comparison of rapid action value estimation variants for general game playing","authors":"C. F. Sironi, M. Winands","doi":"10.1109/CIG.2016.7860429","DOIUrl":"https://doi.org/10.1109/CIG.2016.7860429","url":null,"abstract":"General Game Playing (GGP) aims at creating computer programs able to play any arbitrary game at an expert level given only its rules. The lack of game-specific knowledge and the necessity of learning a strategy online have made Monte-Carlo Tree Search (MCTS) a suitable method to tackle the challenges of GGP. An efficient search-control mechanism can substantially increase the performance of MCTS. The RAVE strategy and its more recent variant, GRAVE, have been proposed for this reason. In this paper we further investigate the use of GRAVE for GGP and compare its performance with the more established RAVE strategy and with a new variant, called HRAVE, that uses more global information. Experiments show that for some games GRAVE and HRAVE perform better than RAVE, with GRAVE being the most promising one overall.","PeriodicalId":6594,"journal":{"name":"2016 IEEE Conference on Computational Intelligence and Games (CIG)","volume":"10 1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2016-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76603339","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 : 2016-09-01DOI: 10.1109/CIG.2016.7860414
Francis Deboeverie, Sanne Roegiers, Gianni Allebosch, P. Veelaert, W. Philips
In this paper, a novel approach for human gesture classification on skeletal data is proposed for the application of exergaming in physiotherapy. Unlike existing methods, we propose to use a general classifier like Random Forests to recognize dynamic gestures. The temporal dimension is handled afterwards by majority voting in a sliding window over the consecutive predictions of the classifier. The gestures can have partially similar postures, such that the classifier will decide on the dissimilar postures. This brute-force classification strategy is permitted, because dynamic human gestures show sufficient dissimilar postures. Online continuous human gesture recognition can classify dynamic gestures in an early stage, which is a crucial advantage when controlling a game by automatic gesture recognition. Also, ground truth can be easily obtained, since all postures in a gesture get the same label, without any discretization into consecutive postures. This way, new gestures can be easily added, which is advantageous in adaptive game development. We evaluate our strategy by a leave-one-subject-out cross-validation on a self-captured stealth game gesture dataset and the publicly available Microsoft Research Cambridge-12 Kinect (MSRC-12) dataset. On the first dataset we achieve an excellent accuracy rate of 96.72%. Furthermore, we show that Random Forests perform better than Support Vector Machines. On the second dataset we achieve an accuracy rate of 98.37%, which is on average 3.57% better then existing methods.
{"title":"Human gesture classification by brute-force machine learning for exergaming in physiotherapy","authors":"Francis Deboeverie, Sanne Roegiers, Gianni Allebosch, P. Veelaert, W. Philips","doi":"10.1109/CIG.2016.7860414","DOIUrl":"https://doi.org/10.1109/CIG.2016.7860414","url":null,"abstract":"In this paper, a novel approach for human gesture classification on skeletal data is proposed for the application of exergaming in physiotherapy. Unlike existing methods, we propose to use a general classifier like Random Forests to recognize dynamic gestures. The temporal dimension is handled afterwards by majority voting in a sliding window over the consecutive predictions of the classifier. The gestures can have partially similar postures, such that the classifier will decide on the dissimilar postures. This brute-force classification strategy is permitted, because dynamic human gestures show sufficient dissimilar postures. Online continuous human gesture recognition can classify dynamic gestures in an early stage, which is a crucial advantage when controlling a game by automatic gesture recognition. Also, ground truth can be easily obtained, since all postures in a gesture get the same label, without any discretization into consecutive postures. This way, new gestures can be easily added, which is advantageous in adaptive game development. We evaluate our strategy by a leave-one-subject-out cross-validation on a self-captured stealth game gesture dataset and the publicly available Microsoft Research Cambridge-12 Kinect (MSRC-12) dataset. On the first dataset we achieve an excellent accuracy rate of 96.72%. Furthermore, we show that Random Forests perform better than Support Vector Machines. On the second dataset we achieve an accuracy rate of 98.37%, which is on average 3.57% better then existing methods.","PeriodicalId":6594,"journal":{"name":"2016 IEEE Conference on Computational Intelligence and Games (CIG)","volume":"92 1","pages":"1-7"},"PeriodicalIF":0.0,"publicationDate":"2016-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73322946","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 : 2016-09-01DOI: 10.1109/CIG.2016.7860447
D. Ventura
This paper argues that computational creativity is the logical next step in the evolution of game design; briefly overviews what is meant by computational creativity and suggests some ways in which it could augment contemporary games; explores some initial ideas for its incorporation into the future of gaming and game design; and argues for increased cross-pollination and collaboration between the computational intelligence and games research community and the computational creativity research community.
{"title":"Beyond computational intelligence to computational creativity in games","authors":"D. Ventura","doi":"10.1109/CIG.2016.7860447","DOIUrl":"https://doi.org/10.1109/CIG.2016.7860447","url":null,"abstract":"This paper argues that computational creativity is the logical next step in the evolution of game design; briefly overviews what is meant by computational creativity and suggests some ways in which it could augment contemporary games; explores some initial ideas for its incorporation into the future of gaming and game design; and argues for increased cross-pollination and collaboration between the computational intelligence and games research community and the computational creativity research community.","PeriodicalId":6594,"journal":{"name":"2016 IEEE Conference on Computational Intelligence and Games (CIG)","volume":"32 1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2016-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78934150","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 : 2016-09-01DOI: 10.1109/CIG.2016.7860394
Santiago Ontañón
The recent success of AlphaGO has shown that it is possible to combine machine learning with Monte Carlo Tree Search (MCTS) in order to improve performance in games with large branching factors. This paper explores the question of whether similar ideas can be applied to a genre of games with an even larger branching factor: Real-Time Strategy games. Specifically, this paper studies (1) the use of Bayesian models to estimate the probability distribution of actions played by a strong player, (2) the incorporation of such models into NaiveMCTS, a MCTS algorithm designed for games with combinatorial branching factors. We call this approach informed MCTS, since it exploits prior information about the game in the form of a probability distribution of actions. We evaluate its performance in the μRTS game simulator, significantly outperforming the previous state of the art.
{"title":"Informed Monte Carlo Tree Search for Real-Time Strategy games","authors":"Santiago Ontañón","doi":"10.1109/CIG.2016.7860394","DOIUrl":"https://doi.org/10.1109/CIG.2016.7860394","url":null,"abstract":"The recent success of AlphaGO has shown that it is possible to combine machine learning with Monte Carlo Tree Search (MCTS) in order to improve performance in games with large branching factors. This paper explores the question of whether similar ideas can be applied to a genre of games with an even larger branching factor: Real-Time Strategy games. Specifically, this paper studies (1) the use of Bayesian models to estimate the probability distribution of actions played by a strong player, (2) the incorporation of such models into NaiveMCTS, a MCTS algorithm designed for games with combinatorial branching factors. We call this approach informed MCTS, since it exploits prior information about the game in the form of a probability distribution of actions. We evaluate its performance in the μRTS game simulator, significantly outperforming the previous state of the art.","PeriodicalId":6594,"journal":{"name":"2016 IEEE Conference on Computational Intelligence and Games (CIG)","volume":"15 1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2016-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81759770","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 : 2016-09-01DOI: 10.1109/CIG.2016.7860440
Yusaku Mandai, Tomoyuki Kaneko
UCT is a standard method of Monte Carlo tree search (MCTS) algorithms, which have been applied to various domains and have achieved remarkable success. This study proposes a family of Leaf-LinUCT, which are improved LinUCT algorithms incorporating LinUCB into MCTS. LinUCB outperforms UCB1 in contextual multi-armed bandit problems, owing to a kind of online learning with ridge regression. However, due to the minimax structure of game trees, ridge regression in LinUCB does not always work well in the context of tree search. In this paper, we remedy the problem and extend our previous work on LinUCT in two ways: (1) by restricting teacher data for regression to the frontier nodes in a current search tree, and (2) by adjusting the feature vector of each internal node to the weighted mean of the feature vector of the descendant nodes. We also present a new synthetic model, incremental-random-feature tree, by extending the standard incremental random tree model. In our model, each node has a feature vector that represents the characteristics of the corresponding position. The elements of a feature vector in a node are randomly changed from those in its parent node by each move, as the heuristic score of a node is randomly changed by each move in the standard incremental random tree model. The experimental results show that our Leaf-LinUCT outperformed UCT and existing LinUCT algorithms, in the incremental-random-feature treeand a synthetic game studied in [1].
{"title":"Improved LinUCT and its evaluation on incremental random-feature tree","authors":"Yusaku Mandai, Tomoyuki Kaneko","doi":"10.1109/CIG.2016.7860440","DOIUrl":"https://doi.org/10.1109/CIG.2016.7860440","url":null,"abstract":"UCT is a standard method of Monte Carlo tree search (MCTS) algorithms, which have been applied to various domains and have achieved remarkable success. This study proposes a family of Leaf-LinUCT, which are improved LinUCT algorithms incorporating LinUCB into MCTS. LinUCB outperforms UCB1 in contextual multi-armed bandit problems, owing to a kind of online learning with ridge regression. However, due to the minimax structure of game trees, ridge regression in LinUCB does not always work well in the context of tree search. In this paper, we remedy the problem and extend our previous work on LinUCT in two ways: (1) by restricting teacher data for regression to the frontier nodes in a current search tree, and (2) by adjusting the feature vector of each internal node to the weighted mean of the feature vector of the descendant nodes. We also present a new synthetic model, incremental-random-feature tree, by extending the standard incremental random tree model. In our model, each node has a feature vector that represents the characteristics of the corresponding position. The elements of a feature vector in a node are randomly changed from those in its parent node by each move, as the heuristic score of a node is randomly changed by each move in the standard incremental random tree model. The experimental results show that our Leaf-LinUCT outperformed UCT and existing LinUCT algorithms, in the incremental-random-feature treeand a synthetic game studied in [1].","PeriodicalId":6594,"journal":{"name":"2016 IEEE Conference on Computational Intelligence and Games (CIG)","volume":"507 1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2016-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86840127","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 : 2016-09-01DOI: 10.1109/CIG.2016.7860445
P. Lopes, Antonios Liapis, Georgios N. Yannakakis
Fear and tension are the primary emotions elicited by the genre of horror, a peculiar characteristic for media whose sole purpose is to entertain. The audience is often lead into tense and fearful situations, meticulously crafted by the authors using a narrative progression and a combination of visual and auditory stimuli. This paper presents a playable demonstration of the Sonancia system, a multi-faceted content generator for 3D horror games, with the capability of generating levels and their corresponding soundscapes. Designers can also guide the level generation process, by defining an intended progression of tension, which the level generator and sonification will adhere to.
{"title":"Sonancia: A multi-faceted generator for horror","authors":"P. Lopes, Antonios Liapis, Georgios N. Yannakakis","doi":"10.1109/CIG.2016.7860445","DOIUrl":"https://doi.org/10.1109/CIG.2016.7860445","url":null,"abstract":"Fear and tension are the primary emotions elicited by the genre of horror, a peculiar characteristic for media whose sole purpose is to entertain. The audience is often lead into tense and fearful situations, meticulously crafted by the authors using a narrative progression and a combination of visual and auditory stimuli. This paper presents a playable demonstration of the Sonancia system, a multi-faceted content generator for 3D horror games, with the capability of generating levels and their corresponding soundscapes. Designers can also guide the level generation process, by defining an intended progression of tension, which the level generator and sonification will adhere to.","PeriodicalId":6594,"journal":{"name":"2016 IEEE Conference on Computational Intelligence and Games (CIG)","volume":"26 1","pages":"1-2"},"PeriodicalIF":0.0,"publicationDate":"2016-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87738760","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 : 2016-09-01DOI: 10.1109/CIG.2016.7860386
Owen Sacco, Antonios Liapis, Georgios N. Yannakakis
The Web contains vast sources of content that could be reused to reduce the development time and effort to create games. However, most Web content is unstructured and lacks meaning for machines to be able to process and infer new knowledge. The Web of Data is a term used to describe a trend for publishing and interlinking previously disconnected datasets on the Web in order to make them more valuable and useful as a whole. In this paper, we describe an innovative approach that exploits Semantic Web technologies to automatically generate games by reusing Web content. Existing work on automatic game content generation through algorithmic means focuses primarily on a set of parameters within constrained game design spaces such as terrains or game levels, but does not harness the potential of already existing content on the Web for game generation. We instead propose a holistic and more generally-applicable game generation solution that would identify suitable Web information sources and enrich game content with semantic meta-structures.
Web包含大量的内容资源，这些内容可以重复使用，从而减少游戏开发的时间和精力。然而，大多数Web内容是非结构化的，缺乏机器处理和推断新知识的意义。Web of Data是一个术语，用于描述一种趋势，即在Web上发布和连接先前断开的数据集，以使它们作为一个整体更有价值和有用。在本文中，我们描述了一种利用语义Web技术通过重用Web内容来自动生成游戏的创新方法。现有的通过算法自动生成游戏内容的工作主要集中在受限的游戏设计空间(如地形或游戏关卡)内的一系列参数，但并没有利用网络上已有内容的潜力来生成游戏。相反，我们提出了一个整体的、更普遍适用的游戏生成解决方案，它将识别合适的Web信息源，并用语义元结构丰富游戏内容。
{"title":"A holistic approach for semantic-based game generation","authors":"Owen Sacco, Antonios Liapis, Georgios N. Yannakakis","doi":"10.1109/CIG.2016.7860386","DOIUrl":"https://doi.org/10.1109/CIG.2016.7860386","url":null,"abstract":"The Web contains vast sources of content that could be reused to reduce the development time and effort to create games. However, most Web content is unstructured and lacks meaning for machines to be able to process and infer new knowledge. The Web of Data is a term used to describe a trend for publishing and interlinking previously disconnected datasets on the Web in order to make them more valuable and useful as a whole. In this paper, we describe an innovative approach that exploits Semantic Web technologies to automatically generate games by reusing Web content. Existing work on automatic game content generation through algorithmic means focuses primarily on a set of parameters within constrained game design spaces such as terrains or game levels, but does not harness the potential of already existing content on the Web for game generation. We instead propose a holistic and more generally-applicable game generation solution that would identify suitable Web information sources and enrich game content with semantic meta-structures.","PeriodicalId":6594,"journal":{"name":"2016 IEEE Conference on Computational Intelligence and Games (CIG)","volume":"28 1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2016-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77049651","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 : 2016-07-08DOI: 10.1109/CIG.2016.7860389
T. Cazenave, Jialin Liu, F. Teytaud, O. Teytaud
Many artificial intelligences (AIs) are randomized. One can be lucky or unlucky with the random seed; we quantify this effect and show that, maybe contrarily to intuition, this is far from being negligible. Then, we apply two different existing algorithms for selecting good seeds and good probability distributions over seeds. This mainly leads to learning an opening book. We apply this to Phantom Go, which, as all phantom games, is hard for opening book learning. We improve the winning rate from 50% to 70% in 5×5 against the same AI, and from approximately 0% to 40% in 5×5, 7×7 and 9×9 against a stronger (learning) opponent.
{"title":"Learning opening books in partially observable games: Using random seeds in Phantom Go","authors":"T. Cazenave, Jialin Liu, F. Teytaud, O. Teytaud","doi":"10.1109/CIG.2016.7860389","DOIUrl":"https://doi.org/10.1109/CIG.2016.7860389","url":null,"abstract":"Many artificial intelligences (AIs) are randomized. One can be lucky or unlucky with the random seed; we quantify this effect and show that, maybe contrarily to intuition, this is far from being negligible. Then, we apply two different existing algorithms for selecting good seeds and good probability distributions over seeds. This mainly leads to learning an opening book. We apply this to Phantom Go, which, as all phantom games, is hard for opening book learning. We improve the winning rate from 50% to 70% in 5×5 against the same AI, and from approximately 0% to 40% in 5×5, 7×7 and 9×9 against a stronger (learning) opponent.","PeriodicalId":6594,"journal":{"name":"2016 IEEE Conference on Computational Intelligence and Games (CIG)","volume":"739 1","pages":"1-7"},"PeriodicalIF":0.0,"publicationDate":"2016-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78386234","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 : 2016-07-02DOI: 10.1109/CIG.2016.7860430
Diego Perez Liebana, Spyridon Samothrakis, J. Togelius, T. Schaul, S. Lucas
This paper presents a study on the robustness and variability of performance of general video game-playing agents. Agents analyzed includes those that won the different legs of the 2014 and 2015 General Video Game AI Competitions, and two sample agents distributed with its framework. Initially, these agents are run in four games and ranked according to the rules of the competition. Then, different modifications to the reward signal of the games are proposed and noise is introduced in either the actions executed by the controller, their forward model, or both. Results show that it is possible to produce a significant change in the rankings by introducing the modifications proposed here. This is an important result because it enables the set of human-authored games to be automatically expanded by adding parameter-varied versions that add information and insight into the relative strengths of the agents under test. Results also show that some controllers perform well under almost all conditions, a testament to the robustness of the GVGAI benchmark.
{"title":"Analyzing the robustness of general video game playing agents","authors":"Diego Perez Liebana, Spyridon Samothrakis, J. Togelius, T. Schaul, S. Lucas","doi":"10.1109/CIG.2016.7860430","DOIUrl":"https://doi.org/10.1109/CIG.2016.7860430","url":null,"abstract":"This paper presents a study on the robustness and variability of performance of general video game-playing agents. Agents analyzed includes those that won the different legs of the 2014 and 2015 General Video Game AI Competitions, and two sample agents distributed with its framework. Initially, these agents are run in four games and ranked according to the rules of the competition. Then, different modifications to the reward signal of the games are proposed and noise is introduced in either the actions executed by the controller, their forward model, or both. Results show that it is possible to produce a significant change in the rankings by introducing the modifications proposed here. This is an important result because it enables the set of human-authored games to be automatically expanded by adding parameter-varied versions that add information and insight into the relative strengths of the agents under test. Results also show that some controllers perform well under almost all conditions, a testament to the robustness of the GVGAI benchmark.","PeriodicalId":6594,"journal":{"name":"2016 IEEE Conference on Computational Intelligence and Games (CIG)","volume":"97 1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2016-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"74351003","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：