Pub Date : 2002-08-07DOI: 10.1109/NAFIPS.2002.1018050
J. M. Adams, K. Rattan
A backpropagation learning method is developed for partitioned, triangular, fuzzy input membership functions to account for the coupled nature of the function parameters. Partitioned, triangular input membership functions are common in industrial fuzzy applications. The resulting algorithm is applied to a Mamdani fuzzy logic system with product-sum inference and weighted-average defuzzification. The algorithm is developed from the standard backpropagation method with the complete impact of each input parameter change included in the partial derivative expansion of the system. The algorithm is applied to tune the input parameters of a controller for a two-link, planar robot. The system response is demonstrated for a set of commands which create cross-coupling through both centrifugal and Coriolis forces.
{"title":"Backpropagation learning for a fuzzy controller with partitioned membership functions","authors":"J. M. Adams, K. Rattan","doi":"10.1109/NAFIPS.2002.1018050","DOIUrl":"https://doi.org/10.1109/NAFIPS.2002.1018050","url":null,"abstract":"A backpropagation learning method is developed for partitioned, triangular, fuzzy input membership functions to account for the coupled nature of the function parameters. Partitioned, triangular input membership functions are common in industrial fuzzy applications. The resulting algorithm is applied to a Mamdani fuzzy logic system with product-sum inference and weighted-average defuzzification. The algorithm is developed from the standard backpropagation method with the complete impact of each input parameter change included in the partial derivative expansion of the system. The algorithm is applied to tune the input parameters of a controller for a two-link, planar robot. The system response is demonstrated for a set of commands which create cross-coupling through both centrifugal and Coriolis forces.","PeriodicalId":348314,"journal":{"name":"2002 Annual Meeting of the North American Fuzzy Information Processing Society Proceedings. NAFIPS-FLINT 2002 (Cat. No. 02TH8622)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129739712","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 : 2002-08-07DOI: 10.1109/NAFIPS.2002.1018106
D. Spiegel, T. Sudkamp
The local evolutionary generation of fuzzy rule bases employs independent searches in local regions throughout the input space and combines the local results to produce a global model. The paper presents a rule base tuning strategy that is compatible with the local evolutionary generation of fuzzy rule bases. Rule base tuning is accomplished by modifying the decomposition of the input domain based on the distribution and values of the training data. A local tuning algorithm must maintain a correspondence between competing rules in the population. An experimental suite has been developed to exhibit the potential for model optimization using rule base tuning. of particular interest is the ability of rule base tuning to compensate for the effects of sparse data.
{"title":"Tuning membership functions in local evolutionary learning of fuzzy rule bases","authors":"D. Spiegel, T. Sudkamp","doi":"10.1109/NAFIPS.2002.1018106","DOIUrl":"https://doi.org/10.1109/NAFIPS.2002.1018106","url":null,"abstract":"The local evolutionary generation of fuzzy rule bases employs independent searches in local regions throughout the input space and combines the local results to produce a global model. The paper presents a rule base tuning strategy that is compatible with the local evolutionary generation of fuzzy rule bases. Rule base tuning is accomplished by modifying the decomposition of the input domain based on the distribution and values of the training data. A local tuning algorithm must maintain a correspondence between competing rules in the population. An experimental suite has been developed to exhibit the potential for model optimization using rule base tuning. of particular interest is the ability of rule base tuning to compensate for the effects of sparse data.","PeriodicalId":348314,"journal":{"name":"2002 Annual Meeting of the North American Fuzzy Information Processing Society Proceedings. NAFIPS-FLINT 2002 (Cat. No. 02TH8622)","volume":"64 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125456314","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 : 2002-08-07DOI: 10.1109/NAFIPS.2002.1018101
L. Mazlack
The paper is an initial exploration of representation, uncertain precision, and dependency as a connected concern that is not easily decomposed. Often for reasons of computational simplicity, the assumption is made that attributes are independent of each other. Similarly, whether dependency exists is classically expressed in terms of first order logic. The actuality is that often the assumptions produce results that are not good representations of reality. An integrated representation of multiple, related attributes is difficult. Usually, different attributes have different ranges and linearity. Sometimes, normalization is a first step in meaningfully representing different kinds of data, or, as a first step to the combination different kinds of data. Most values are not crisply known. A method of dependency representation is needed. The issue of how to represent both dependency and uncertainty needs to be resolved. In a very real sense, dependency representations and their imprecision both enriches and constrains how we approach the solutions to our problems.
{"title":"Representation, uncertain imprecision, and dependency","authors":"L. Mazlack","doi":"10.1109/NAFIPS.2002.1018101","DOIUrl":"https://doi.org/10.1109/NAFIPS.2002.1018101","url":null,"abstract":"The paper is an initial exploration of representation, uncertain precision, and dependency as a connected concern that is not easily decomposed. Often for reasons of computational simplicity, the assumption is made that attributes are independent of each other. Similarly, whether dependency exists is classically expressed in terms of first order logic. The actuality is that often the assumptions produce results that are not good representations of reality. An integrated representation of multiple, related attributes is difficult. Usually, different attributes have different ranges and linearity. Sometimes, normalization is a first step in meaningfully representing different kinds of data, or, as a first step to the combination different kinds of data. Most values are not crisply known. A method of dependency representation is needed. The issue of how to represent both dependency and uncertainty needs to be resolved. In a very real sense, dependency representations and their imprecision both enriches and constrains how we approach the solutions to our problems.","PeriodicalId":348314,"journal":{"name":"2002 Annual Meeting of the North American Fuzzy Information Processing Society Proceedings. NAFIPS-FLINT 2002 (Cat. No. 02TH8622)","volume":"70 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124130995","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 : 2002-08-07DOI: 10.1109/NAFIPS.2002.1018064
S. Mccaslin, A. Kulkarni
In this paper, we propose a method to extract and reduce fuzzy IF-THEN rules from fuzzy neural systems. After training, fuzzy rules are extracted from the fuzzy neural network by backtracking along the weighted paths through the neural network. These rules are then reduced by use of a fuzzy associative memory (FAM) bank. We used this algorithm to extract classification rules from a multi-spectral satellite image. The image represents the Mississippi river bottomland. In order to verify the rule extraction method, measures such as accuracy, overall Kappa and fidelity are used. The results are presented in the paper.
{"title":"Knowledge discovery from multispectral satellite images","authors":"S. Mccaslin, A. Kulkarni","doi":"10.1109/NAFIPS.2002.1018064","DOIUrl":"https://doi.org/10.1109/NAFIPS.2002.1018064","url":null,"abstract":"In this paper, we propose a method to extract and reduce fuzzy IF-THEN rules from fuzzy neural systems. After training, fuzzy rules are extracted from the fuzzy neural network by backtracking along the weighted paths through the neural network. These rules are then reduced by use of a fuzzy associative memory (FAM) bank. We used this algorithm to extract classification rules from a multi-spectral satellite image. The image represents the Mississippi river bottomland. In order to verify the rule extraction method, measures such as accuracy, overall Kappa and fidelity are used. The results are presented in the paper.","PeriodicalId":348314,"journal":{"name":"2002 Annual Meeting of the North American Fuzzy Information Processing Society Proceedings. NAFIPS-FLINT 2002 (Cat. No. 02TH8622)","volume":"289 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125665939","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 : 2002-08-07DOI: 10.1109/NAFIPS.2002.1018063
O. Sjahputera, P. Matsakis, J. Keller, R. Bondugula
In ongoing work on spatial relations and scene interpretation, we present a system that linguistically describes the motion of an object in a temporal sequence. This description, called dynamic linguistic description, is inferred from a sequence of static linguistic descriptions explaining the relative position, at different instances, between a moving object and a stationary object. In this preliminary work, the moving object is assumed to be moving in a straight path at a constant velocity. The scene is monitored from a fixed pose with a constant frame rate. The proposed system is potentially useful as a low-bandwidth remote observation system capable of linguistically reporting relative position and motion in a scene.
{"title":"Linguistic descriptions for an object in motion","authors":"O. Sjahputera, P. Matsakis, J. Keller, R. Bondugula","doi":"10.1109/NAFIPS.2002.1018063","DOIUrl":"https://doi.org/10.1109/NAFIPS.2002.1018063","url":null,"abstract":"In ongoing work on spatial relations and scene interpretation, we present a system that linguistically describes the motion of an object in a temporal sequence. This description, called dynamic linguistic description, is inferred from a sequence of static linguistic descriptions explaining the relative position, at different instances, between a moving object and a stationary object. In this preliminary work, the moving object is assumed to be moving in a straight path at a constant velocity. The scene is monitored from a fixed pose with a constant frame rate. The proposed system is potentially useful as a low-bandwidth remote observation system capable of linguistically reporting relative position and motion in a scene.","PeriodicalId":348314,"journal":{"name":"2002 Annual Meeting of the North American Fuzzy Information Processing Society Proceedings. NAFIPS-FLINT 2002 (Cat. No. 02TH8622)","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133611990","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 : 2002-08-07DOI: 10.1109/NAFIPS.2002.1018030
M. Sato-Ilic
Proposes an estimation method for fuzzy cluster loading using the kernel method. Fuzzy cluster loading was proposed in order to interpret the result of fuzzy clustering by obtaining the relationship between the obtained fuzzy clusters and the variables of the given data. From the structure of the model for fuzzy cluster loading, it is known that the estimate is obtained using the estimate of the weighted regression analysis. We propose a method to obtain the estimate in a higher space then the space in the given data using the idea of the kernel method. The significant properties of this technique are: (1) we use high dimension space to estimate the fuzzy cluster loading, due to this, we can get a better result to extract the data structure; and (2) through the cluster structure of given data, we can extract a clearer structure of the given data. Several numerical examples show the validity of the proposed technique and the efficiency of the use of the cluster structure in the given data.
{"title":"Fuzzy regression analysis using fuzzy clustering","authors":"M. Sato-Ilic","doi":"10.1109/NAFIPS.2002.1018030","DOIUrl":"https://doi.org/10.1109/NAFIPS.2002.1018030","url":null,"abstract":"Proposes an estimation method for fuzzy cluster loading using the kernel method. Fuzzy cluster loading was proposed in order to interpret the result of fuzzy clustering by obtaining the relationship between the obtained fuzzy clusters and the variables of the given data. From the structure of the model for fuzzy cluster loading, it is known that the estimate is obtained using the estimate of the weighted regression analysis. We propose a method to obtain the estimate in a higher space then the space in the given data using the idea of the kernel method. The significant properties of this technique are: (1) we use high dimension space to estimate the fuzzy cluster loading, due to this, we can get a better result to extract the data structure; and (2) through the cluster structure of given data, we can extract a clearer structure of the given data. Several numerical examples show the validity of the proposed technique and the efficiency of the use of the cluster structure in the given data.","PeriodicalId":348314,"journal":{"name":"2002 Annual Meeting of the North American Fuzzy Information Processing Society Proceedings. NAFIPS-FLINT 2002 (Cat. No. 02TH8622)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133419891","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 : 2002-08-07DOI: 10.1109/NAFIPS.2002.1018127
Dimitar Filev
This paper deals with an intelligent system approach to the problem of control of plants with large parameters variations and multiple operating modes. It is based on the concept of a dynamic model bank - a long-term memory working in conjunction with a recursive learning algorithm that online estimates plant dynamics. The role of the model bank is to accumulate these models that successfully approximate the plant and to further use them to improve the performance of an indirect adaptive control algorithm. The models contained in the bank are used to periodically initialize a recursive least-square estimation procedure in the cases when it cannot provide a satisfactory approximation of the plant. An OWA aggregation operator that is dependent on the performance of individual models is applied to infer the initializing model parameters. The bank is continually updated by summarizing the parameters of the estimated models without requirement for off-line identification.
{"title":"Model bank based intelligent control","authors":"Dimitar Filev","doi":"10.1109/NAFIPS.2002.1018127","DOIUrl":"https://doi.org/10.1109/NAFIPS.2002.1018127","url":null,"abstract":"This paper deals with an intelligent system approach to the problem of control of plants with large parameters variations and multiple operating modes. It is based on the concept of a dynamic model bank - a long-term memory working in conjunction with a recursive learning algorithm that online estimates plant dynamics. The role of the model bank is to accumulate these models that successfully approximate the plant and to further use them to improve the performance of an indirect adaptive control algorithm. The models contained in the bank are used to periodically initialize a recursive least-square estimation procedure in the cases when it cannot provide a satisfactory approximation of the plant. An OWA aggregation operator that is dependent on the performance of individual models is applied to infer the initializing model parameters. The bank is continually updated by summarizing the parameters of the estimated models without requirement for off-line identification.","PeriodicalId":348314,"journal":{"name":"2002 Annual Meeting of the North American Fuzzy Information Processing Society Proceedings. NAFIPS-FLINT 2002 (Cat. No. 02TH8622)","volume":"34 5","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131790609","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 : 2002-08-07DOI: 10.1109/NAFIPS.2002.1018065
C. García‐Barroso, P. Sobrevilla, A.L. Pedra, E. Montseny
This paper presents an edge detection algorithm whose output provides, for each pixel, the degree to which it looks like an edge pixel. The algorithm design is based on the analysis of gradient vectors' magnitudes and directions within a local neighborhood of the pixel being considered. The use of fuzzy techniques makes it possible for the algorithm to be parameter-independent for adapting it to the image to be analyzed. Moreover, since it is based on a local analysis, it can be processed in parallel, allowing real-time execution.
{"title":"Fuzzy contour detection based on a good approximation of the argument of the gradient vector","authors":"C. García‐Barroso, P. Sobrevilla, A.L. Pedra, E. Montseny","doi":"10.1109/NAFIPS.2002.1018065","DOIUrl":"https://doi.org/10.1109/NAFIPS.2002.1018065","url":null,"abstract":"This paper presents an edge detection algorithm whose output provides, for each pixel, the degree to which it looks like an edge pixel. The algorithm design is based on the analysis of gradient vectors' magnitudes and directions within a local neighborhood of the pixel being considered. The use of fuzzy techniques makes it possible for the algorithm to be parameter-independent for adapting it to the image to be analyzed. Moreover, since it is based on a local analysis, it can be processed in parallel, allowing real-time execution.","PeriodicalId":348314,"journal":{"name":"2002 Annual Meeting of the North American Fuzzy Information Processing Society Proceedings. NAFIPS-FLINT 2002 (Cat. No. 02TH8622)","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131908343","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 : 2002-08-07DOI: 10.1109/NAFIPS.2002.1018071
Wen-Ran Zhang
It is observed that bipolar equilibrium is truth. Bipolarity like fuzziness, as an integral part of truth, is inseparable from the truth. It is argued that Boolean logic and its extensions cannot be directly used to represent and reason with the coexistence of bipolar truth. A crisp bipolar combinational logic BCL/sub I/ and its fuzzy counterpart BCL/sub F/ are proposed for bipolar modeling based on the ancient Chinese Yin-Yang philosophy. The new family of logical systems is distinguished from other models. Bipolar zero-order completeness and 1st-order soundness are proved with a Hilbert style proof theory based on unipolar axioms and rules of inference. In addition, the lair's case in the ancient paradox is redressed.
{"title":"Bipolar logic and bipolar fuzzy logic - a unification of truth, fuzziness, and bipolarity","authors":"Wen-Ran Zhang","doi":"10.1109/NAFIPS.2002.1018071","DOIUrl":"https://doi.org/10.1109/NAFIPS.2002.1018071","url":null,"abstract":"It is observed that bipolar equilibrium is truth. Bipolarity like fuzziness, as an integral part of truth, is inseparable from the truth. It is argued that Boolean logic and its extensions cannot be directly used to represent and reason with the coexistence of bipolar truth. A crisp bipolar combinational logic BCL/sub I/ and its fuzzy counterpart BCL/sub F/ are proposed for bipolar modeling based on the ancient Chinese Yin-Yang philosophy. The new family of logical systems is distinguished from other models. Bipolar zero-order completeness and 1st-order soundness are proved with a Hilbert style proof theory based on unipolar axioms and rules of inference. In addition, the lair's case in the ancient paradox is redressed.","PeriodicalId":348314,"journal":{"name":"2002 Annual Meeting of the North American Fuzzy Information Processing Society Proceedings. NAFIPS-FLINT 2002 (Cat. No. 02TH8622)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131208642","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 : 2002-08-07DOI: 10.1109/NAFIPS.2002.1018025
I. Turksen, A. Esper, K. Patel, S. Starks, V. Kreinovich
In classical (two-valued) logic, CNF and DNF forms of each propositional formula are equivalent to each other. In fuzzy logic, CNF and DNF forms are not equivalent, they form an interval that contains the fuzzy values of all classically equivalent propositional formulas. If we want to select a single value from this interval, then it is natural to select a linear combination of the interval's endpoints. In particular, we can do that for CNF and DNF forms of "and" and "or", thus designing natural fuzzy analogues of classical "and" and "or" operations. The problem with thus selected "and" and "or" operations is that, contrary to common sense expectations, they are not associative. We show the largest possible value of the corresponding non-associativity is reasonably small and thus, this non-associativity does not make these operations impractical.
{"title":"Selecting a fuzzy logic operation from the DNF-CNF interval: how practical are the resulting operations?","authors":"I. Turksen, A. Esper, K. Patel, S. Starks, V. Kreinovich","doi":"10.1109/NAFIPS.2002.1018025","DOIUrl":"https://doi.org/10.1109/NAFIPS.2002.1018025","url":null,"abstract":"In classical (two-valued) logic, CNF and DNF forms of each propositional formula are equivalent to each other. In fuzzy logic, CNF and DNF forms are not equivalent, they form an interval that contains the fuzzy values of all classically equivalent propositional formulas. If we want to select a single value from this interval, then it is natural to select a linear combination of the interval's endpoints. In particular, we can do that for CNF and DNF forms of \"and\" and \"or\", thus designing natural fuzzy analogues of classical \"and\" and \"or\" operations. The problem with thus selected \"and\" and \"or\" operations is that, contrary to common sense expectations, they are not associative. We show the largest possible value of the corresponding non-associativity is reasonably small and thus, this non-associativity does not make these operations impractical.","PeriodicalId":348314,"journal":{"name":"2002 Annual Meeting of the North American Fuzzy Information Processing Society Proceedings. NAFIPS-FLINT 2002 (Cat. No. 02TH8622)","volume":"40 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124998904","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：