Pub Date : 2003-07-24DOI: 10.1109/NAFIPS.2003.1226826
D. Iourinski, François Modave
In multicriteria decision making (MCDM), we aim at ranking multidimensional alternatives. A traditional approach utility-like approach is to define an appropriate aggregation operator, that aggregates partial preferences based on the decision maker's behavior. Non-additive measures (or fuzzy measures) have been shown to be well-suited tools for this purpose. However, this was done in an ad hoc way until recently. An axiomatization of multicriteria decision making was given in a quantitative setting, using the Choquet integral for aggregation operator. However, this choice of fuzzy integral is not always adequate from a measurement perspective. The aim of this paper is to give conditions for the existence of a Sugeno integral with respect to some fuzzy measure, representing the global preferences of a decision maker.
{"title":"Qualitative multicriteria decision making based on the Sugeno integral","authors":"D. Iourinski, François Modave","doi":"10.1109/NAFIPS.2003.1226826","DOIUrl":"https://doi.org/10.1109/NAFIPS.2003.1226826","url":null,"abstract":"In multicriteria decision making (MCDM), we aim at ranking multidimensional alternatives. A traditional approach utility-like approach is to define an appropriate aggregation operator, that aggregates partial preferences based on the decision maker's behavior. Non-additive measures (or fuzzy measures) have been shown to be well-suited tools for this purpose. However, this was done in an ad hoc way until recently. An axiomatization of multicriteria decision making was given in a quantitative setting, using the Choquet integral for aggregation operator. However, this choice of fuzzy integral is not always adequate from a measurement perspective. The aim of this paper is to give conditions for the existence of a Sugeno integral with respect to some fuzzy measure, representing the global preferences of a decision maker.","PeriodicalId":153530,"journal":{"name":"22nd International Conference of the North American Fuzzy Information Processing Society, NAFIPS 2003","volume":"172 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133751456","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 : 2003-07-24DOI: 10.1109/NAFIPS.2003.1226834
P. Bonissone
We describe the life cycle of a fuzzy knowledge-based classifier with special emphasis on one of its most neglected steps: the maintenance of its knowledge base. First, we analyze the process of underwriting Insurance applications, which is the classification problem used to illustrate the life cycle of a classifier. After discussing some design tradeoffs that must be addressed for the on-line and off-line use of a classifier, we describe the design and implementation of a fuzzy rule-based (FRB) and a fuzzy case-based (FCB) classifier. We establish a standard reference dataset (SRD), consisting of 3,000 insurance applications with their corresponding decisions. The SRD exemplifies the results achieved by an ideal, optimal classifier, and represents the target for our design. We apply evolutionary algorithms to perform an off-line optimization of the design parameters of each classifier, modifying their behavior to approximate this target. The SRD is also used as a reference for testing and performing a five-fold cross-validation of the classifiers. Finally, we focus on the monitoring and maintenance of the FRB classifier. We describe a fusion architecture that supports an off-line quality assurance process of the on-line FRB classifier. The fusion module takes the outputs of multiple classifiers, determines their degree of consensus, and compares their overall agreement with that of the FRB classifier. From this analysis, we can identify the most suitable cases to update the SRD, to audit, or to be reviewed by senior underwriters.
{"title":"The life cycle of a fuzzy knowledge-based classifier","authors":"P. Bonissone","doi":"10.1109/NAFIPS.2003.1226834","DOIUrl":"https://doi.org/10.1109/NAFIPS.2003.1226834","url":null,"abstract":"We describe the life cycle of a fuzzy knowledge-based classifier with special emphasis on one of its most neglected steps: the maintenance of its knowledge base. First, we analyze the process of underwriting Insurance applications, which is the classification problem used to illustrate the life cycle of a classifier. After discussing some design tradeoffs that must be addressed for the on-line and off-line use of a classifier, we describe the design and implementation of a fuzzy rule-based (FRB) and a fuzzy case-based (FCB) classifier. We establish a standard reference dataset (SRD), consisting of 3,000 insurance applications with their corresponding decisions. The SRD exemplifies the results achieved by an ideal, optimal classifier, and represents the target for our design. We apply evolutionary algorithms to perform an off-line optimization of the design parameters of each classifier, modifying their behavior to approximate this target. The SRD is also used as a reference for testing and performing a five-fold cross-validation of the classifiers. Finally, we focus on the monitoring and maintenance of the FRB classifier. We describe a fusion architecture that supports an off-line quality assurance process of the on-line FRB classifier. The fusion module takes the outputs of multiple classifiers, determines their degree of consensus, and compares their overall agreement with that of the FRB classifier. From this analysis, we can identify the most suitable cases to update the SRD, to audit, or to be reviewed by senior underwriters.","PeriodicalId":153530,"journal":{"name":"22nd International Conference of the North American Fuzzy Information Processing Society, NAFIPS 2003","volume":"148 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123775336","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 : 2003-07-24DOI: 10.1109/NAFIPS.2003.1226750
J. M. Barone, P. Dewan
While studies of the role of fuzzy logic in natural language certainly exist, it is not clear that the use of fuzzy logic to represent linguistic constructs is anything more than an engineering convenience. This paper suggests that one reason this situation obtains is because fuzzy logic has been used strictly to elucidate static aspects of natural language (particularly aspects of the lexicon). If one examines dynamic features of natural language, on the other hand, new possibilities for connections between fuzzy logic and natural language emerge. In particular, some results from category theory are used to show that fuzzy logic can have a role in explaining certain otherwise rather obscure properties of linguistic comparatives in English.
{"title":"Looking for fuzziness in natural language","authors":"J. M. Barone, P. Dewan","doi":"10.1109/NAFIPS.2003.1226750","DOIUrl":"https://doi.org/10.1109/NAFIPS.2003.1226750","url":null,"abstract":"While studies of the role of fuzzy logic in natural language certainly exist, it is not clear that the use of fuzzy logic to represent linguistic constructs is anything more than an engineering convenience. This paper suggests that one reason this situation obtains is because fuzzy logic has been used strictly to elucidate static aspects of natural language (particularly aspects of the lexicon). If one examines dynamic features of natural language, on the other hand, new possibilities for connections between fuzzy logic and natural language emerge. In particular, some results from category theory are used to show that fuzzy logic can have a role in explaining certain otherwise rather obscure properties of linguistic comparatives in English.","PeriodicalId":153530,"journal":{"name":"22nd International Conference of the North American Fuzzy Information Processing Society, NAFIPS 2003","volume":"138 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127862322","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 : 2003-07-24DOI: 10.1109/NAFIPS.2003.1226830
A. Klimke
The transformation method has been proposed for the simulation and analysis of systems with uncertain parameters. Here, several aspects of an efficient implementation are presented: fast processing of discretized fuzzy numbers through multi-dimensional arrays, elimination of recurring permutations, automatic decomposition of models, treatment of single occurrences of variables through interval arithmetic, and a monotonicity test based on automatic differentiation.
{"title":"An efficient implementation of the transformation method of fuzzy arithmetic","authors":"A. Klimke","doi":"10.1109/NAFIPS.2003.1226830","DOIUrl":"https://doi.org/10.1109/NAFIPS.2003.1226830","url":null,"abstract":"The transformation method has been proposed for the simulation and analysis of systems with uncertain parameters. Here, several aspects of an efficient implementation are presented: fast processing of discretized fuzzy numbers through multi-dimensional arrays, elimination of recurring permutations, automatic decomposition of models, treatment of single occurrences of variables through interval arithmetic, and a monotonicity test based on automatic differentiation.","PeriodicalId":153530,"journal":{"name":"22nd International Conference of the North American Fuzzy Information Processing Society, NAFIPS 2003","volume":"293 2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121264977","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 : 2003-07-24DOI: 10.1109/NAFIPS.2003.1226827
W.U. Syed
Computational models of the bargaining process at a farmers' market require modeling subjective preferences of the buyers and sellers and their subjective assessments of the produce. The proposed fuzzy agents based model employs a number of fuzzy inference systems modeled as Standard Additive Models that perform the subjective decision making for the agents. Survey results of different vendors and customers at different farmers' markets provide the rule base coded in these fuzzy expert systems. The results show a steady convergence to a bargain weighted towards the greedier of the two players. Repeated simulations with the proposed model of varying buyers, sellers and the produce indicate that fuzzy agents can model bargaining at a farmers' market.
{"title":"Fuzzy agents bargaining at a farmer's market","authors":"W.U. Syed","doi":"10.1109/NAFIPS.2003.1226827","DOIUrl":"https://doi.org/10.1109/NAFIPS.2003.1226827","url":null,"abstract":"Computational models of the bargaining process at a farmers' market require modeling subjective preferences of the buyers and sellers and their subjective assessments of the produce. The proposed fuzzy agents based model employs a number of fuzzy inference systems modeled as Standard Additive Models that perform the subjective decision making for the agents. Survey results of different vendors and customers at different farmers' markets provide the rule base coded in these fuzzy expert systems. The results show a steady convergence to a bargain weighted towards the greedier of the two players. Repeated simulations with the proposed model of varying buyers, sellers and the produce indicate that fuzzy agents can model bargaining at a farmers' market.","PeriodicalId":153530,"journal":{"name":"22nd International Conference of the North American Fuzzy Information Processing Society, NAFIPS 2003","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126837616","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 : 2003-07-24DOI: 10.1109/NAFIPS.2003.1226748
J. Paetz
The starting point for this contribution is an adapted neuro-fuzzy system of Huber/Berthold with a set of adapted membership functions (number and shape). The heuristically adapted number and shape of the membership functions may not be the best choice, especially when considering human understandability of the adapted rules. We transform a-posteriori the number of fuzzy terms and evaluate classification performance and understandability, considering the influence of the weighting of the neuro-fuzzy units as well. Inference for the new, transformed (deduced) system is done by an expanded max-min inference strategy. For this expanded inference the influence of the neuro-fuzzy membership functions to the predefined number of fuzzy terms have to be determined. Thus, we introduce so called degradation factors. The evaluation of our inventions is done by medical data.
{"title":"Deducing fuzzy inference systems with different numbers of membership functions from a neuro-fuzzy inference system","authors":"J. Paetz","doi":"10.1109/NAFIPS.2003.1226748","DOIUrl":"https://doi.org/10.1109/NAFIPS.2003.1226748","url":null,"abstract":"The starting point for this contribution is an adapted neuro-fuzzy system of Huber/Berthold with a set of adapted membership functions (number and shape). The heuristically adapted number and shape of the membership functions may not be the best choice, especially when considering human understandability of the adapted rules. We transform a-posteriori the number of fuzzy terms and evaluate classification performance and understandability, considering the influence of the weighting of the neuro-fuzzy units as well. Inference for the new, transformed (deduced) system is done by an expanded max-min inference strategy. For this expanded inference the influence of the neuro-fuzzy membership functions to the predefined number of fuzzy terms have to be determined. Thus, we introduce so called degradation factors. The evaluation of our inventions is done by medical data.","PeriodicalId":153530,"journal":{"name":"22nd International Conference of the North American Fuzzy Information Processing Society, NAFIPS 2003","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115765090","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 : 2003-07-24DOI: 10.1109/NAFIPS.2003.1226819
Jianzhong Zhang, D. Berleant
A cumulative distribution function (CDF) states the probability that a sample of a random variable will be no greater than a value x, where x is a real value. Closed form expressions for important CDFs have parameters, such as mean and variance. If these parameters are not point values but rather intervals, sharp or fuzzy, then a single CDF is not specified. Instead, a family of CDFs is specified. Sharp intervals lead to sharp boundaries ("envelopes") around the family, while fuzzy intervals lead to fuzzy boundaries. Algorithms exist that compute the family of CDFs possible for some function g(v) where v is a vector of distributions or bounded families of distribution. We investigate the bounds on families of CDFs implied by interval values for their parameters. These bounds can then be used as inputs to algorithms that manipulate distributions and bounded spaces defining families of distributions (sometimes called probability boxes or p-boxes). For example, problems defining inputs this way may be found in. In this paper, we present the bounds for the families of a few common CDFs when parameters to those CDFs are intervals.
{"title":"Envelopes around cumulative distribution functions from interval parameters of standard continuous distributions","authors":"Jianzhong Zhang, D. Berleant","doi":"10.1109/NAFIPS.2003.1226819","DOIUrl":"https://doi.org/10.1109/NAFIPS.2003.1226819","url":null,"abstract":"A cumulative distribution function (CDF) states the probability that a sample of a random variable will be no greater than a value x, where x is a real value. Closed form expressions for important CDFs have parameters, such as mean and variance. If these parameters are not point values but rather intervals, sharp or fuzzy, then a single CDF is not specified. Instead, a family of CDFs is specified. Sharp intervals lead to sharp boundaries (\"envelopes\") around the family, while fuzzy intervals lead to fuzzy boundaries. Algorithms exist that compute the family of CDFs possible for some function g(v) where v is a vector of distributions or bounded families of distribution. We investigate the bounds on families of CDFs implied by interval values for their parameters. These bounds can then be used as inputs to algorithms that manipulate distributions and bounded spaces defining families of distributions (sometimes called probability boxes or p-boxes). For example, problems defining inputs this way may be found in. In this paper, we present the bounds for the families of a few common CDFs when parameters to those CDFs are intervals.","PeriodicalId":153530,"journal":{"name":"22nd International Conference of the North American Fuzzy Information Processing Society, NAFIPS 2003","volume":"94 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130076413","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 : 2003-07-24DOI: 10.1109/NAFIPS.2003.1226795
JingTao Yao
An important issue of data mining is how to transfer data into information, the information into action, and the action into value or profit. This paper presents a study on applying sensitivity analysis to neural network models for a particular area in data mining, interesting mining and profit mining. Applying sensitivity analysis to neural network models rather than just regression models can help us identify sensible factors that play important roles to dependent variables such as total profit in a dynamic environment.
{"title":"Sensitivity analysis for data mining","authors":"JingTao Yao","doi":"10.1109/NAFIPS.2003.1226795","DOIUrl":"https://doi.org/10.1109/NAFIPS.2003.1226795","url":null,"abstract":"An important issue of data mining is how to transfer data into information, the information into action, and the action into value or profit. This paper presents a study on applying sensitivity analysis to neural network models for a particular area in data mining, interesting mining and profit mining. Applying sensitivity analysis to neural network models rather than just regression models can help us identify sensible factors that play important roles to dependent variables such as total profit in a dynamic environment.","PeriodicalId":153530,"journal":{"name":"22nd International Conference of the North American Fuzzy Information Processing Society, NAFIPS 2003","volume":"5 3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122544646","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 : 2003-07-24DOI: 10.1109/NAFIPS.2003.1226774
P. Bosc, O. Pivert
This paper is situated in the area of possibilistic relational databases, i.e., where some attribute values are imprecise and represented as possibility distributions. Any such database has a canonical interpretation as a set of regular relational databases, called worlds. This view provides the basic semantics of any query addressed to a possibilistic database. However, a query cannot be run this way for tractability reasons. This situation has led us to consider specific families of queries that can be processed in a compact way, i.e., directly on possibilistic relations. The queries dealt with in this paper, called necessity-based queries, are of the form: "to what extent is it certain that tuple t belongs to the result of query Q", where Q denotes a regular relational query. The major contribution of this paper is to identify the constraints over Q (in terms of algebraic operations) which must be imposed so that these queries are tractable.
{"title":"About certainty-based queries against possibilistic databases","authors":"P. Bosc, O. Pivert","doi":"10.1109/NAFIPS.2003.1226774","DOIUrl":"https://doi.org/10.1109/NAFIPS.2003.1226774","url":null,"abstract":"This paper is situated in the area of possibilistic relational databases, i.e., where some attribute values are imprecise and represented as possibility distributions. Any such database has a canonical interpretation as a set of regular relational databases, called worlds. This view provides the basic semantics of any query addressed to a possibilistic database. However, a query cannot be run this way for tractability reasons. This situation has led us to consider specific families of queries that can be processed in a compact way, i.e., directly on possibilistic relations. The queries dealt with in this paper, called necessity-based queries, are of the form: \"to what extent is it certain that tuple t belongs to the result of query Q\", where Q denotes a regular relational query. The major contribution of this paper is to identify the constraints over Q (in terms of algebraic operations) which must be imposed so that these queries are tractable.","PeriodicalId":153530,"journal":{"name":"22nd International Conference of the North American Fuzzy Information Processing Society, NAFIPS 2003","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122786694","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 : 2003-07-24DOI: 10.1109/NAFIPS.2003.1226776
R. Thomopoulos, P. Bosc, P. Buche, O. Haemmerle
In previous studies, we have extended the conceptual graph model, which is a knowledge representation model belonging to the family of semantic networks, to be able to represent fuzzy values. The basic conceptual graph model has a logical interpretation in first-order logic. In this paper, we focus on the logical interpretation of the conceptual graph model extended to fuzzy values: we use logical implications stemming from fuzzy logic, so as to extend the logical interpretation of the model to fuzzy values and to comparisons between fuzzy conceptual graphs.
{"title":"Logical interpretations of fuzzy conceptual graphs","authors":"R. Thomopoulos, P. Bosc, P. Buche, O. Haemmerle","doi":"10.1109/NAFIPS.2003.1226776","DOIUrl":"https://doi.org/10.1109/NAFIPS.2003.1226776","url":null,"abstract":"In previous studies, we have extended the conceptual graph model, which is a knowledge representation model belonging to the family of semantic networks, to be able to represent fuzzy values. The basic conceptual graph model has a logical interpretation in first-order logic. In this paper, we focus on the logical interpretation of the conceptual graph model extended to fuzzy values: we use logical implications stemming from fuzzy logic, so as to extend the logical interpretation of the model to fuzzy values and to comparisons between fuzzy conceptual graphs.","PeriodicalId":153530,"journal":{"name":"22nd International Conference of the North American Fuzzy Information Processing Society, NAFIPS 2003","volume":"31 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117035748","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：