Pub Date : 2001-06-25DOI: 10.1109/SMCIA.2001.936727
W. Land, T. Masters, J. Lo, D.W. McKee, F. R. Anderson
A new neural network technology was developed to improve the diagnosis of breast cancer using mammogram findings. The paradigm, adaptive boosting (AB), uses a markedly different theory in solving the computational intelligence (CI) problem. AB, a new machine learning paradigm, focuses on finding weak learning algorithm(s) that initially need to provide slightly better than "random" performance (i.e., approximately 55%) when processing a mammogram training set. By successive development of additional architectures (using the mammogram training set), the adaptive boosting process improves performance of the basic evolutionary programming derived neural network architectures. The results of these several EP-derived hybrid architectures are then intelligently combined and tested using a similar validation mammogram data set. Optimization, focused on improving specificity and positive predictive value at very high sensitivities, with an analysis of the performance of the hybrid would be most meaningful. Using the DUKE mammogram database of 500 biopsy proven samples, this hybrid, on average, was able to achieve (under statistical 5-fold cross-validation) a specificity of 48.3% and a positive predictive value (PPV) of 51.8% while maintaining 100% sensitivity. At 97% sensitivity, a specificity of 56.6% and a PPV of 55.8% were obtained.
{"title":"New results in breast cancer classification obtained from an evolutionary computation/adaptive boosting hybrid using mammogram and history data","authors":"W. Land, T. Masters, J. Lo, D.W. McKee, F. R. Anderson","doi":"10.1109/SMCIA.2001.936727","DOIUrl":"https://doi.org/10.1109/SMCIA.2001.936727","url":null,"abstract":"A new neural network technology was developed to improve the diagnosis of breast cancer using mammogram findings. The paradigm, adaptive boosting (AB), uses a markedly different theory in solving the computational intelligence (CI) problem. AB, a new machine learning paradigm, focuses on finding weak learning algorithm(s) that initially need to provide slightly better than \"random\" performance (i.e., approximately 55%) when processing a mammogram training set. By successive development of additional architectures (using the mammogram training set), the adaptive boosting process improves performance of the basic evolutionary programming derived neural network architectures. The results of these several EP-derived hybrid architectures are then intelligently combined and tested using a similar validation mammogram data set. Optimization, focused on improving specificity and positive predictive value at very high sensitivities, with an analysis of the performance of the hybrid would be most meaningful. Using the DUKE mammogram database of 500 biopsy proven samples, this hybrid, on average, was able to achieve (under statistical 5-fold cross-validation) a specificity of 48.3% and a positive predictive value (PPV) of 51.8% while maintaining 100% sensitivity. At 97% sensitivity, a specificity of 56.6% and a PPV of 55.8% were obtained.","PeriodicalId":104202,"journal":{"name":"SMCia/01. Proceedings of the 2001 IEEE Mountain Workshop on Soft Computing in Industrial Applications (Cat. No.01EX504)","volume":"37 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2001-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133147047","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 : 2001-06-25DOI: 10.1109/SMCIA.2001.936723
H. Lewis
The paper presents a system for day-ahead load forecasting as originally proposed to a regional electric power company. The company provided funding for developing most parts of this software. The system is based on a hybrid approach to intelligent systems design combining a fuzzy heuristic approach based on the knowledge of human experts in load forecasting with a data-driven neural network-based component. To make the system truly useful, considerable emphasis was placed on the user interface including a highly developed explanation module.
{"title":"Intelligent hybrid load forecasting system for an electric power company","authors":"H. Lewis","doi":"10.1109/SMCIA.2001.936723","DOIUrl":"https://doi.org/10.1109/SMCIA.2001.936723","url":null,"abstract":"The paper presents a system for day-ahead load forecasting as originally proposed to a regional electric power company. The company provided funding for developing most parts of this software. The system is based on a hybrid approach to intelligent systems design combining a fuzzy heuristic approach based on the knowledge of human experts in load forecasting with a data-driven neural network-based component. To make the system truly useful, considerable emphasis was placed on the user interface including a highly developed explanation module.","PeriodicalId":104202,"journal":{"name":"SMCia/01. Proceedings of the 2001 IEEE Mountain Workshop on Soft Computing in Industrial Applications (Cat. No.01EX504)","volume":"84 15","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2001-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131472333","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 : 2001-06-25DOI: 10.1109/SMCIA.2001.936744
X.Z. Gao, S. Ovaska, Y. Dote
We propose a new dynamical fuzzy system with linguistic information feedback (DFSLIF). Instead of crisp system output, the delayed conclusion fuzzy membership function in the consequence part is fed back locally with adjustable scaling and shifting in order to overcome the static mapping drawback of conventional fuzzy systems. We give a detailed description of the corresponding structure and algorithm. Our novel scheme has the advantage of inherent dynamics, and is therefore well suited for handling temporal problems like dynamical system identification, control, and filtering. Simulation experiments have been carried out to demonstrate its effectiveness.
{"title":"DFSLIF: dynamical fuzzy system with linguistic information feedback","authors":"X.Z. Gao, S. Ovaska, Y. Dote","doi":"10.1109/SMCIA.2001.936744","DOIUrl":"https://doi.org/10.1109/SMCIA.2001.936744","url":null,"abstract":"We propose a new dynamical fuzzy system with linguistic information feedback (DFSLIF). Instead of crisp system output, the delayed conclusion fuzzy membership function in the consequence part is fed back locally with adjustable scaling and shifting in order to overcome the static mapping drawback of conventional fuzzy systems. We give a detailed description of the corresponding structure and algorithm. Our novel scheme has the advantage of inherent dynamics, and is therefore well suited for handling temporal problems like dynamical system identification, control, and filtering. Simulation experiments have been carried out to demonstrate its effectiveness.","PeriodicalId":104202,"journal":{"name":"SMCia/01. Proceedings of the 2001 IEEE Mountain Workshop on Soft Computing in Industrial Applications (Cat. No.01EX504)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2001-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131748625","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 : 2001-06-25DOI: 10.1109/SMCIA.2001.936728
M. Ozdemir, M. Embrechts, F. Arciniegas, C. Breneman, L. Lockwood, Kristin P. Bennett
QSAR (quantitative structure activity relationship) is a discipline within computational chemistry that deals with predictive modeling, often for relatively small datasets where the number of features might exceed the number of data points, leading to extreme dimensionality problems. The paper addresses a novel feature selection procedure for QSAR based on genetic algorithms to reduce the curse of dimensionality problem. In this case the genetic algorithm minimizes a cost function derived from the correlation matrix between the features and the activity of interest that is being modeled. From a QSAR dataset with 160 features, the genetic algorithm selected a feature subset (40 features), which built a better predictive model than with full feature set. The results for feature reduction with genetic algorithm were also compared with neural network sensitivity analysis.
{"title":"Feature selection for in-silico drug design using genetic algorithms and neural networks","authors":"M. Ozdemir, M. Embrechts, F. Arciniegas, C. Breneman, L. Lockwood, Kristin P. Bennett","doi":"10.1109/SMCIA.2001.936728","DOIUrl":"https://doi.org/10.1109/SMCIA.2001.936728","url":null,"abstract":"QSAR (quantitative structure activity relationship) is a discipline within computational chemistry that deals with predictive modeling, often for relatively small datasets where the number of features might exceed the number of data points, leading to extreme dimensionality problems. The paper addresses a novel feature selection procedure for QSAR based on genetic algorithms to reduce the curse of dimensionality problem. In this case the genetic algorithm minimizes a cost function derived from the correlation matrix between the features and the activity of interest that is being modeled. From a QSAR dataset with 160 features, the genetic algorithm selected a feature subset (40 features), which built a better predictive model than with full feature set. The results for feature reduction with genetic algorithm were also compared with neural network sensitivity analysis.","PeriodicalId":104202,"journal":{"name":"SMCia/01. Proceedings of the 2001 IEEE Mountain Workshop on Soft Computing in Industrial Applications (Cat. No.01EX504)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2001-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130866264","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 : 2001-06-25DOI: 10.1109/SMCIA.2001.936720
H. Vanlandingham, F. Azam, W. Pulliam
The paper presents two methods of training multilayer perceptrons (MLPs) that use both functional values and co-located derivative values during the training process. The first method extends the standard backpropagation training algorithm for MLPs whereas the second method employs genetic algorithms (GAs) to find the optimal neural network weights using both functional and co-located function derivative values. The GAs used for optimization of the weights of a feedforward artificial neural network use a special reordering of the genotype before recombination. The ultimate goal of this research effort is to be able to train and design an artificial neural networks (ANN) more effectively, i.e., to have a network that generalizes better, learns faster and requires fewer training data points. The initial results indicate that the methods do, in fact, provide good generalization while requiring only a relatively sparse sampling of the function and its derivative values during the training phase, as indicated by the illustrative examples.
{"title":"Neural network training for complex industrial applications","authors":"H. Vanlandingham, F. Azam, W. Pulliam","doi":"10.1109/SMCIA.2001.936720","DOIUrl":"https://doi.org/10.1109/SMCIA.2001.936720","url":null,"abstract":"The paper presents two methods of training multilayer perceptrons (MLPs) that use both functional values and co-located derivative values during the training process. The first method extends the standard backpropagation training algorithm for MLPs whereas the second method employs genetic algorithms (GAs) to find the optimal neural network weights using both functional and co-located function derivative values. The GAs used for optimization of the weights of a feedforward artificial neural network use a special reordering of the genotype before recombination. The ultimate goal of this research effort is to be able to train and design an artificial neural networks (ANN) more effectively, i.e., to have a network that generalizes better, learns faster and requires fewer training data points. The initial results indicate that the methods do, in fact, provide good generalization while requiring only a relatively sparse sampling of the function and its derivative values during the training phase, as indicated by the illustrative examples.","PeriodicalId":104202,"journal":{"name":"SMCia/01. Proceedings of the 2001 IEEE Mountain Workshop on Soft Computing in Industrial Applications (Cat. No.01EX504)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2001-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128313073","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 : 2001-06-25DOI: 10.1109/SMCIA.2001.936721
M. Embrechts, F. Arciniegas, M. Ozdemir, M. Momma
The paper introduces scientific data mining, the standard data-mining problem, and the strip-mining problem. StripMiner/sup TM/, a shell program for feature reduction and predictive modeling, integrates the executions of several different machine-learning models (partial least squares regression, genetic algorithms, support vector machines, neural networks, and local learning). This paper introduces the StripMiner/sup TM/ code, its functionality, and its options.
{"title":"Scientific data mining with StripMiner/sup TM/","authors":"M. Embrechts, F. Arciniegas, M. Ozdemir, M. Momma","doi":"10.1109/SMCIA.2001.936721","DOIUrl":"https://doi.org/10.1109/SMCIA.2001.936721","url":null,"abstract":"The paper introduces scientific data mining, the standard data-mining problem, and the strip-mining problem. StripMiner/sup TM/, a shell program for feature reduction and predictive modeling, integrates the executions of several different machine-learning models (partial least squares regression, genetic algorithms, support vector machines, neural networks, and local learning). This paper introduces the StripMiner/sup TM/ code, its functionality, and its options.","PeriodicalId":104202,"journal":{"name":"SMCia/01. Proceedings of the 2001 IEEE Mountain Workshop on Soft Computing in Industrial Applications (Cat. No.01EX504)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2001-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114988908","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 : 2001-06-25DOI: 10.1109/SMCIA.2001.936746
H. Takagi
We first overview the research trend of computational intelligence, discuss what comes next in the computational intelligence research, and conclude that humanized technologies would be one of the essential keywords of the possible research direction. Then, we take up interactive evolutionary computation (IEC) as one of the humanized technologies and show how IEC technology has spread to a wide variety of fields, what problems remain, and what kinds of challenges need to be solved, and how to make the technology practical.
{"title":"Advancing the human experience with interactive evolutionary computation","authors":"H. Takagi","doi":"10.1109/SMCIA.2001.936746","DOIUrl":"https://doi.org/10.1109/SMCIA.2001.936746","url":null,"abstract":"We first overview the research trend of computational intelligence, discuss what comes next in the computational intelligence research, and conclude that humanized technologies would be one of the essential keywords of the possible research direction. Then, we take up interactive evolutionary computation (IEC) as one of the humanized technologies and show how IEC technology has spread to a wide variety of fields, what problems remain, and what kinds of challenges need to be solved, and how to make the technology practical.","PeriodicalId":104202,"journal":{"name":"SMCia/01. Proceedings of the 2001 IEEE Mountain Workshop on Soft Computing in Industrial Applications (Cat. No.01EX504)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2001-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122621432","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 : 2001-06-25DOI: 10.1109/SMCIA.2001.936740
D. Teodorovic, P. Lucic
The problem considered in this paper is as follows: assign the available aircraft from a fleet to specific routes so that the aircraft are kept in normal operation as long as possible before going to the technical base, taking care that "higher quality" aircraft are assigned to "more important" routes. This paper develops a model for aircraft assignment that includes both numerical and linguistic information normally used by dispatchers. The developed model is tested on a real numerical example.
{"title":"Approximate reasoning algorithm for short term aircraft assignment","authors":"D. Teodorovic, P. Lucic","doi":"10.1109/SMCIA.2001.936740","DOIUrl":"https://doi.org/10.1109/SMCIA.2001.936740","url":null,"abstract":"The problem considered in this paper is as follows: assign the available aircraft from a fleet to specific routes so that the aircraft are kept in normal operation as long as possible before going to the technical base, taking care that \"higher quality\" aircraft are assigned to \"more important\" routes. This paper develops a model for aircraft assignment that includes both numerical and linguistic information normally used by dispatchers. The developed model is tested on a real numerical example.","PeriodicalId":104202,"journal":{"name":"SMCia/01. Proceedings of the 2001 IEEE Mountain Workshop on Soft Computing in Industrial Applications (Cat. No.01EX504)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2001-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129061499","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 : 2001-06-25DOI: 10.1109/SMCIA.2001.936736
P. Dadone, H. Vanlandingham
This paper proposes and describes an application of remote control for an industrial process. A control program implemented on a PC using the Java language allows for easy prototyping of a fuzzy logic controller. The PC is connected through a data acquisition card to a laboratory process (used in a control teaching laboratory). The fuzzy controller is easily setup and adjusted, allowing for the control of the process. The Java implementation conceptually permits a portable and remote measurement and control approach for any industrial process.
{"title":"Remote control of industrial processes","authors":"P. Dadone, H. Vanlandingham","doi":"10.1109/SMCIA.2001.936736","DOIUrl":"https://doi.org/10.1109/SMCIA.2001.936736","url":null,"abstract":"This paper proposes and describes an application of remote control for an industrial process. A control program implemented on a PC using the Java language allows for easy prototyping of a fuzzy logic controller. The PC is connected through a data acquisition card to a laboratory process (used in a control teaching laboratory). The fuzzy controller is easily setup and adjusted, allowing for the control of the process. The Java implementation conceptually permits a portable and remote measurement and control approach for any industrial process.","PeriodicalId":104202,"journal":{"name":"SMCia/01. Proceedings of the 2001 IEEE Mountain Workshop on Soft Computing in Industrial Applications (Cat. No.01EX504)","volume":"69 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2001-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129079170","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 : 2001-06-25DOI: 10.1109/SMCIA.2001.936724
A. Kamiya, K. Kawai, I. Ono, S. Kobayashi
Power plant start-up scheduling is aimed at minimizing the start-up time while limiting maximum turbine-rotor stresses. A shorter start-up time not only reduces fuel and electricity consumption during the start-up process, but also increases its capability of adapting to changes in electricity demand. This scheduling problem is, however, highly nonlinear with a number of local optima within a wide search space. In our previous research, we proved that the optimal schedule stays on the edge of the feasible space, and provided an adaptive enforcement operation based on a theoretical setting equation. The adaptive enforcement operation used with GA is applied to compel the search along the edge of the feasible space, so as to increase the search efficiency. We give a brief description of the theoretical proof and present simulation test results with a range of hard-to-search stress limit sets to verify the search efficiency of the theoretically-proved search model.
发电厂启动调度的目的是在限制涡轮机转子最大应力的同时，尽量缩短启动时间。缩短启动时间不仅能减少启动过程中的燃料和电力消耗，还能提高适应电力需求变化的能力。然而，这个调度问题是一个高度非线性的问题，在广阔的搜索空间内存在许多局部最优。在之前的研究中，我们证明了最优调度停留在可行空间的边缘，并提供了基于理论设定方程的自适应执行操作。与 GA 配合使用的自适应执行操作用于强制沿可行空间边缘搜索，从而提高搜索效率。我们简要介绍了理论证明，并给出了一系列难搜索压力极限集的仿真测试结果，以验证理论证明的搜索模型的搜索效率。
{"title":"Adaptive power plant start-up scheduling: simulation test results","authors":"A. Kamiya, K. Kawai, I. Ono, S. Kobayashi","doi":"10.1109/SMCIA.2001.936724","DOIUrl":"https://doi.org/10.1109/SMCIA.2001.936724","url":null,"abstract":"Power plant start-up scheduling is aimed at minimizing the start-up time while limiting maximum turbine-rotor stresses. A shorter start-up time not only reduces fuel and electricity consumption during the start-up process, but also increases its capability of adapting to changes in electricity demand. This scheduling problem is, however, highly nonlinear with a number of local optima within a wide search space. In our previous research, we proved that the optimal schedule stays on the edge of the feasible space, and provided an adaptive enforcement operation based on a theoretical setting equation. The adaptive enforcement operation used with GA is applied to compel the search along the edge of the feasible space, so as to increase the search efficiency. We give a brief description of the theoretical proof and present simulation test results with a range of hard-to-search stress limit sets to verify the search efficiency of the theoretically-proved search model.","PeriodicalId":104202,"journal":{"name":"SMCia/01. Proceedings of the 2001 IEEE Mountain Workshop on Soft Computing in Industrial Applications (Cat. No.01EX504)","volume":"26 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2001-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132057669","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：