Pub Date : 2008-03-04DOI: 10.1109/GEFS.2008.4484555
Yaochu Jin
This talk attempts to argue that almost all adaptive systems have multiple objectives to achieve. Very often, there is no single solution that can optimize all objectives, in which case, the concept of Pareto-optimization plays an important rule. Examples will be given ranging from engineering design, machine learning, to biological systems to show how Pareto-optimality can make a difference in analyzing these systems. The first example we will discuss is the aerodynamic design optimization of turbine blades, where energy efficiency in terms of pressure loss as well as the variation of pressure distribution must be minimized. One additional difficulty in aerodynamic design optimization is that the quality of candidate designs must be assessed by performing computational fluid dynamics analysis, which is very time consuming. To reduce computation time, computational techniques like parallel computation, and machine learning techniques, such as meta-modeling can be employed.Surprisingly interesting results will also be achieved when the concept of Pareto-optimality is applied to machine learning. Two cases will be provided to illustrate this idea. In the first case, we show how Pareto-based approach can address neural network regularization more elegantly, through which deeper insights into the problem can be gained. In the second case, we show that analysis of the Pareto-optimal solutions will help determine the optimal number of clusters in data clustering, which again shown how the Pareto front can disclose additional knowledge about the problem at hand. The final example is concerned with tradeoffs in simulated evolution of genetic representation. It has been argued that robustness is critical for biological evolution, because without certain degree of robustness to mutations, it is impossible for evolution to create new functionalities. Therefore, evolution must find representations that are sufficiently robust yet have the potential to innovate. Examples will be given to show that such tradeoff does exist in evolving both a stationary genotype-phenotype mapping, and also a gene regulatory network described by a random Boolean network.
{"title":"Pareto-optimality is everywhere: From engineering design, machine learning, to biological systems","authors":"Yaochu Jin","doi":"10.1109/GEFS.2008.4484555","DOIUrl":"https://doi.org/10.1109/GEFS.2008.4484555","url":null,"abstract":"This talk attempts to argue that almost all adaptive systems have multiple objectives to achieve. Very often, there is no single solution that can optimize all objectives, in which case, the concept of Pareto-optimization plays an important rule. Examples will be given ranging from engineering design, machine learning, to biological systems to show how Pareto-optimality can make a difference in analyzing these systems. The first example we will discuss is the aerodynamic design optimization of turbine blades, where energy efficiency in terms of pressure loss as well as the variation of pressure distribution must be minimized. One additional difficulty in aerodynamic design optimization is that the quality of candidate designs must be assessed by performing computational fluid dynamics analysis, which is very time consuming. To reduce computation time, computational techniques like parallel computation, and machine learning techniques, such as meta-modeling can be employed.Surprisingly interesting results will also be achieved when the concept of Pareto-optimality is applied to machine learning. Two cases will be provided to illustrate this idea. In the first case, we show how Pareto-based approach can address neural network regularization more elegantly, through which deeper insights into the problem can be gained. In the second case, we show that analysis of the Pareto-optimal solutions will help determine the optimal number of clusters in data clustering, which again shown how the Pareto front can disclose additional knowledge about the problem at hand. The final example is concerned with tradeoffs in simulated evolution of genetic representation. It has been argued that robustness is critical for biological evolution, because without certain degree of robustness to mutations, it is impossible for evolution to create new functionalities. Therefore, evolution must find representations that are sufficiently robust yet have the potential to innovate. Examples will be given to show that such tradeoff does exist in evolving both a stationary genotype-phenotype mapping, and also a gene regulatory network described by a random Boolean network.","PeriodicalId":297294,"journal":{"name":"IEEE Workshop on Genetic and Evolutionary Fuzzy Systems","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-03-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130326020","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 : 2008-03-04DOI: 10.1109/GEFS.2008.4484557
W. Pedrycz
In this study, we augment the highly impressive record of developments of fuzzy models by bringing the ideas of evolvable and knowledge-consistent fuzzy modeling. More often than in the past, we are exposed to highly distributed data reflecting some temporal or spatial variability of the problem. Owing to some non-technical reasons (e.g., data privacy and security) or existing technical constraints, the models built locally cannot take advantage of the data available elsewhere. Instead one could be provided with some more abstract entities such as information granules that are reflective of the knowledge conveyed by some other models which could be effectively shared. The two main categories of design schemes discussed here demonstrate the effect of achieving knowledge consistency which augments the existing paradigm of fuzzy modeling. In the first one, we are concerned with sharing temporal knowledge where the models are formed for temporal data available in successive time slices pertinent to the problem at hand and the available temporal knowledge (captured in terms of the structure and parameters of the models) whose usage incorporates the factor of time. In this sense, the resulting fuzzy models become highly evolvable modeling architectures. The spatial nature of knowledge is associated with fuzzy models which are constructed on a basis of data pertinent to some local regions (such as sections of wireless sensor networks, sales regions, etc.). While the introduced conceptual developments are of substantial level of generality, the study will focus on a family of rule-based fuzzy models to illustrate the ensuing algorithmic aspects of the fundamental concepts.
{"title":"Concepts of evolvable and knowledge-consistent fuzzy models","authors":"W. Pedrycz","doi":"10.1109/GEFS.2008.4484557","DOIUrl":"https://doi.org/10.1109/GEFS.2008.4484557","url":null,"abstract":"In this study, we augment the highly impressive record of developments of fuzzy models by bringing the ideas of evolvable and knowledge-consistent fuzzy modeling. More often than in the past, we are exposed to highly distributed data reflecting some temporal or spatial variability of the problem. Owing to some non-technical reasons (e.g., data privacy and security) or existing technical constraints, the models built locally cannot take advantage of the data available elsewhere. Instead one could be provided with some more abstract entities such as information granules that are reflective of the knowledge conveyed by some other models which could be effectively shared. The two main categories of design schemes discussed here demonstrate the effect of achieving knowledge consistency which augments the existing paradigm of fuzzy modeling. In the first one, we are concerned with sharing temporal knowledge where the models are formed for temporal data available in successive time slices pertinent to the problem at hand and the available temporal knowledge (captured in terms of the structure and parameters of the models) whose usage incorporates the factor of time. In this sense, the resulting fuzzy models become highly evolvable modeling architectures. The spatial nature of knowledge is associated with fuzzy models which are constructed on a basis of data pertinent to some local regions (such as sections of wireless sensor networks, sales regions, etc.). While the introduced conceptual developments are of substantial level of generality, the study will focus on a family of rule-based fuzzy models to illustrate the ensuing algorithmic aspects of the fundamental concepts.","PeriodicalId":297294,"journal":{"name":"IEEE Workshop on Genetic and Evolutionary Fuzzy Systems","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-03-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133439796","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：