Reinforcement learning, while being a highly popular learning technique for agents and multi-agent systems, has so far encountered difficulties when applying it to more complex domains due to scaling-up problems. This paper focuses on the use of domain knowledge to improve the convergence speed and optimality of various RL techniques. Specifically, we propose the use of high-level STRIPS operator knowledge in reward shaping to focus the search for the optimal policy. Empirical results show that the plan-based reward shaping approach outperforms other RL techniques, including alternative manual and MDP-based reward shaping when it is used in its basic form. We show that MDP-based reward shaping may fail and successful experiments with STRIPS-based shaping suggest modifications which can overcome encountered problems. The STRIPS-based method we propose allows expressing the same domain knowledge in a different way and the domain expert can choose whether to define an MDP or STRIPS planning task. We also evaluate the robustness of the proposed STRIPS-based technique to errors in the plan knowledge.
{"title":"Plan-based reward shaping for reinforcement learning","authors":"M. Grzes, D. Kudenko","doi":"10.1109/IS.2008.4670492","DOIUrl":"https://doi.org/10.1109/IS.2008.4670492","url":null,"abstract":"Reinforcement learning, while being a highly popular learning technique for agents and multi-agent systems, has so far encountered difficulties when applying it to more complex domains due to scaling-up problems. This paper focuses on the use of domain knowledge to improve the convergence speed and optimality of various RL techniques. Specifically, we propose the use of high-level STRIPS operator knowledge in reward shaping to focus the search for the optimal policy. Empirical results show that the plan-based reward shaping approach outperforms other RL techniques, including alternative manual and MDP-based reward shaping when it is used in its basic form. We show that MDP-based reward shaping may fail and successful experiments with STRIPS-based shaping suggest modifications which can overcome encountered problems. The STRIPS-based method we propose allows expressing the same domain knowledge in a different way and the domain expert can choose whether to define an MDP or STRIPS planning task. We also evaluate the robustness of the proposed STRIPS-based technique to errors in the plan knowledge.","PeriodicalId":305750,"journal":{"name":"2008 4th International IEEE Conference Intelligent Systems","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124895008","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}
P. Papazoglou, Dimitrios Alexios Karras, R. C. Papademetriou
Channel allocation in wireless communication systems is one of the fundamental issues. The corresponding allocation schemes can not be static due to the dynamically changing traffic conditions and network performance. Thus, more sophisticated strategies adapted to current network conditions must be investigated and applied. Recently, various approaches have been proposed for channel allocation based on intelligent techniques such as multi-agent technology and genetic algorithms. These approaches constitute heuristic solutions to resource management problem. On the other hand, the ant colony optimization approach has been proposed for solving optimization problems but this approach has not been proposed so far for solving the channel allocation problem in wireless communication systems. In this paper, a comprehensive heuristic approach for solving the channel allocation problem based on intelligent techniques such as multi-agents and ant colony optimization is proposed. Moreover, important implementation issues such as thread execution sequence are also presented. Finally, the simulation results show the performance improvement of the proposed ant colony optimization algorithm as well as the multi-agent modeling approach.
{"title":"On the implementation of ant colony optimization scheme for improved channel allocation in wireless communications","authors":"P. Papazoglou, Dimitrios Alexios Karras, R. C. Papademetriou","doi":"10.1109/IS.2008.4670437","DOIUrl":"https://doi.org/10.1109/IS.2008.4670437","url":null,"abstract":"Channel allocation in wireless communication systems is one of the fundamental issues. The corresponding allocation schemes can not be static due to the dynamically changing traffic conditions and network performance. Thus, more sophisticated strategies adapted to current network conditions must be investigated and applied. Recently, various approaches have been proposed for channel allocation based on intelligent techniques such as multi-agent technology and genetic algorithms. These approaches constitute heuristic solutions to resource management problem. On the other hand, the ant colony optimization approach has been proposed for solving optimization problems but this approach has not been proposed so far for solving the channel allocation problem in wireless communication systems. In this paper, a comprehensive heuristic approach for solving the channel allocation problem based on intelligent techniques such as multi-agents and ant colony optimization is proposed. Moreover, important implementation issues such as thread execution sequence are also presented. Finally, the simulation results show the performance improvement of the proposed ant colony optimization algorithm as well as the multi-agent modeling approach.","PeriodicalId":305750,"journal":{"name":"2008 4th International IEEE Conference Intelligent Systems","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131779647","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}
The objective of this paper is to propose a framework of ERP systems implementation success that helps project managers to deal with large and complex ERP systems implementation projects. The framework is designed to evaluate and to select activities or alternatives to improvement the degree of performance by the success factors by using the analytic network process (ANP) and the priority matrix (PM). The experience with the application of ANP in a framework of ERP systems implementation success suggests that it offers guidelines in the evaluation and selection of alternatives.
{"title":"Application of the Analytic Network Process (ANP) in a framework of ERP systems implementation success","authors":"V. Dimitrova","doi":"10.1109/IS.2008.4670468","DOIUrl":"https://doi.org/10.1109/IS.2008.4670468","url":null,"abstract":"The objective of this paper is to propose a framework of ERP systems implementation success that helps project managers to deal with large and complex ERP systems implementation projects. The framework is designed to evaluate and to select activities or alternatives to improvement the degree of performance by the success factors by using the analytic network process (ANP) and the priority matrix (PM). The experience with the application of ANP in a framework of ERP systems implementation success suggests that it offers guidelines in the evaluation and selection of alternatives.","PeriodicalId":305750,"journal":{"name":"2008 4th International IEEE Conference Intelligent Systems","volume":"160 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132633944","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}
The paper intends to present a modeling philosophy, that allows a global view of a generic manufacturing enterprise behavior, combining managerial and control approaches. The main objective of the assessment is to design a system architecture for manufacturing that will be able to develop an intelligent comportment displaying characteristics like environment adaptability, problem solving and learning from experience. The foundations of this approach are mainly knowledge management and control engineering techniques, as well as agent-based architectures.
{"title":"Towards a new generation of intelligent manufacturing systems","authors":"I. Dumitrache, S. Caramihai","doi":"10.1109/IS.2008.4670422","DOIUrl":"https://doi.org/10.1109/IS.2008.4670422","url":null,"abstract":"The paper intends to present a modeling philosophy, that allows a global view of a generic manufacturing enterprise behavior, combining managerial and control approaches. The main objective of the assessment is to design a system architecture for manufacturing that will be able to develop an intelligent comportment displaying characteristics like environment adaptability, problem solving and learning from experience. The foundations of this approach are mainly knowledge management and control engineering techniques, as well as agent-based architectures.","PeriodicalId":305750,"journal":{"name":"2008 4th International IEEE Conference Intelligent Systems","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133431425","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}
Analysis of contemporary information security systems (ISS) and especially the case of intrusion detection systems (IDS) shows us few character negative features and drawbacks. Original methods and combined anomaly and signature IDS applications are presented in the paper. Human-centered methods INCONSISTENCY, FUNNEL, CALEIDOSCOPE and CROSSWORD interact on a competitive principle and are controlled by a synthetic metamethod SMM. A research is going on for the possibilities of including other machine learning and data mining methods under the general control of SMM. Their applications aim at computational discovery and knowledge acquisition. It is reinforced by identification and resolution of contradictions, self-learning and other methods for analysis of different types of models from the ISS domain. The complexity of application results is considered. The data analysis in the field frequently needs an act of creation especially if it is applied in a knowledge-poor environment. It is shown that even in this case the creative processes are based on applications of clear and well-formalized methods.
{"title":"Novel intrusion prevention and detection methods","authors":"V. Jotsov","doi":"10.1109/IS.2008.4670526","DOIUrl":"https://doi.org/10.1109/IS.2008.4670526","url":null,"abstract":"Analysis of contemporary information security systems (ISS) and especially the case of intrusion detection systems (IDS) shows us few character negative features and drawbacks. Original methods and combined anomaly and signature IDS applications are presented in the paper. Human-centered methods INCONSISTENCY, FUNNEL, CALEIDOSCOPE and CROSSWORD interact on a competitive principle and are controlled by a synthetic metamethod SMM. A research is going on for the possibilities of including other machine learning and data mining methods under the general control of SMM. Their applications aim at computational discovery and knowledge acquisition. It is reinforced by identification and resolution of contradictions, self-learning and other methods for analysis of different types of models from the ISS domain. The complexity of application results is considered. The data analysis in the field frequently needs an act of creation especially if it is applied in a knowledge-poor environment. It is shown that even in this case the creative processes are based on applications of clear and well-formalized methods.","PeriodicalId":305750,"journal":{"name":"2008 4th International IEEE Conference Intelligent Systems","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122212987","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}
The family of Ordered Weighted Averaging (OWA) operators, as introduced by Yager, appear to be very useful in multi-criteria decision-making (MCDM) and It provides a general class of parameterized aggregation operators that include the min, max, average operators. In this paper, we present Analog CMOS implementation of OWA operator in current-mode and 0.35 mum CMOS technologies for the first time. The proposed circuit has high flexibility and programmability and accuracy .We simulated our proposed design in 49 level HSPICE simulation software and its results show that this block can be used in analog fuzzy logic controller chip (FLC).
由Yager介绍的有序加权平均(OWA)算子族在多准则决策(MCDM)中非常有用，它提供了一类一般的参数化聚合算子，包括最小、最大、平均算子。在本文中，我们首次提出了OWA算子在电流模式下的模拟CMOS实现和0.35 μ m CMOS技术。该电路具有较高的灵活性、可编程性和精度，并在49级HSPICE仿真软件中进行了仿真，结果表明该电路可用于模拟模糊控制器芯片(FLC)。
{"title":"Analog CMOS implementation of order weight average operator for fuzzy logic controller chip","authors":"K. Gheysari, A. Khoei, K. Hadidi, M. Mokarram","doi":"10.1109/IS.2008.4670401","DOIUrl":"https://doi.org/10.1109/IS.2008.4670401","url":null,"abstract":"The family of Ordered Weighted Averaging (OWA) operators, as introduced by Yager, appear to be very useful in multi-criteria decision-making (MCDM) and It provides a general class of parameterized aggregation operators that include the min, max, average operators. In this paper, we present Analog CMOS implementation of OWA operator in current-mode and 0.35 mum CMOS technologies for the first time. The proposed circuit has high flexibility and programmability and accuracy .We simulated our proposed design in 49 level HSPICE simulation software and its results show that this block can be used in analog fuzzy logic controller chip (FLC).","PeriodicalId":305750,"journal":{"name":"2008 4th International IEEE Conference Intelligent Systems","volume":"112 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117255654","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}
This paper considers the problem of finding facial features in frontal color images by offering a method for efficient detection. For a given image, our method firstly detects the skin regions on by color segmentation. This segmentation is performed in a 2D projection of the selected color space (HSV, YES, YCrCb or normalized RGB). After that face candidates are found. Then we apply fuzzy mathematical morphology methods to locate precisely the right eyebrow (if exists). Thus we obtain a meaningful feature. A computer implementation in MATLAB is made for experimentation. The visual results of the experiments show the plausibility of proposed techniques.
{"title":"A new approach for finding face features in color images","authors":"A. Popov, D. Dimitrova","doi":"10.1109/IS.2008.4670517","DOIUrl":"https://doi.org/10.1109/IS.2008.4670517","url":null,"abstract":"This paper considers the problem of finding facial features in frontal color images by offering a method for efficient detection. For a given image, our method firstly detects the skin regions on by color segmentation. This segmentation is performed in a 2D projection of the selected color space (HSV, YES, YCrCb or normalized RGB). After that face candidates are found. Then we apply fuzzy mathematical morphology methods to locate precisely the right eyebrow (if exists). Thus we obtain a meaningful feature. A computer implementation in MATLAB is made for experimentation. The visual results of the experiments show the plausibility of proposed techniques.","PeriodicalId":305750,"journal":{"name":"2008 4th International IEEE Conference Intelligent Systems","volume":"132 2-4","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120918033","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}
The paper considers new techniques for clustering and sorting objects that are described with many quantitative and/or qualitative attributes and may exist in several copies with inconsistent and contradictory attributes. Methods are based on the theory of multiset metric spaces. New options for the objectspsila aggregation and features of classes generated are discussed.
{"title":"Clustering and sorting multi-attribute objects in multiset metric space","authors":"A. Petrovsky","doi":"10.1109/IS.2008.4670507","DOIUrl":"https://doi.org/10.1109/IS.2008.4670507","url":null,"abstract":"The paper considers new techniques for clustering and sorting objects that are described with many quantitative and/or qualitative attributes and may exist in several copies with inconsistent and contradictory attributes. Methods are based on the theory of multiset metric spaces. New options for the objectspsila aggregation and features of classes generated are discussed.","PeriodicalId":305750,"journal":{"name":"2008 4th International IEEE Conference Intelligent Systems","volume":"58 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114639730","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}
Information understanding and information valuation have mostly been studied separately. In this paper we propose using information understanding to reach a definition for information value or quality. We demonstrate how information valuation can be predicted and computed as potential causal impact of extractable semantics.
{"title":"Semantics-based information valuation","authors":"Sinan Al-Saffar, G. Heileman","doi":"10.1109/IS.2008.4670438","DOIUrl":"https://doi.org/10.1109/IS.2008.4670438","url":null,"abstract":"Information understanding and information valuation have mostly been studied separately. In this paper we propose using information understanding to reach a definition for information value or quality. We demonstrate how information valuation can be predicted and computed as potential causal impact of extractable semantics.","PeriodicalId":305750,"journal":{"name":"2008 4th International IEEE Conference Intelligent Systems","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114760972","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}
Mathematical definition of a class dynamic systems with entropy operator is considered (DSEO) The main properties of the entropy operator - existing, continuity, boundedness, differentiability are investigated. The stability conditions are obtained for the DSEO with multiplicative and additive flows. The theoretical results are used for mathematical modeling of labor market and in dynamic procedure of the image restoration from projections.
{"title":"Mathematical models of dynamic systems with entropy operator and their applications","authors":"Y. Popkov","doi":"10.1109/IS.2008.4670395","DOIUrl":"https://doi.org/10.1109/IS.2008.4670395","url":null,"abstract":"Mathematical definition of a class dynamic systems with entropy operator is considered (DSEO) The main properties of the entropy operator - existing, continuity, boundedness, differentiability are investigated. The stability conditions are obtained for the DSEO with multiplicative and additive flows. The theoretical results are used for mathematical modeling of labor market and in dynamic procedure of the image restoration from projections.","PeriodicalId":305750,"journal":{"name":"2008 4th International IEEE Conference Intelligent Systems","volume":"111 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122117454","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：