Pub Date : 1994-03-01DOI: 10.1109/CAIA.1994.323686
Changhwan Lee
Similarity-based learning has been widely and successfully used in some domains. Despite these successes, most similarity measures used in the current literature are defined on limited feature types. Therefore, these similarity measures cannot be applied to the database environment due to the variety of data types that exist. In this paper, we propose a new method of similarity-based learning for databases using information theory. The current similarity measures are improved in several ways. Similarity is defined on every attribute type in the database, and each attribute is assigned a weight depending on its importance with respect to the target attribute. Besides, our nearest neighbor algorithm gives different weights to the selected instances. Our system is implemented and tested on some typical machine learning databases. Our experiments show that the classification accuracy of our system is, in general, superior to that of other learning methods.<>
{"title":"An information theoretic similarity-based learning method for databases","authors":"Changhwan Lee","doi":"10.1109/CAIA.1994.323686","DOIUrl":"https://doi.org/10.1109/CAIA.1994.323686","url":null,"abstract":"Similarity-based learning has been widely and successfully used in some domains. Despite these successes, most similarity measures used in the current literature are defined on limited feature types. Therefore, these similarity measures cannot be applied to the database environment due to the variety of data types that exist. In this paper, we propose a new method of similarity-based learning for databases using information theory. The current similarity measures are improved in several ways. Similarity is defined on every attribute type in the database, and each attribute is assigned a weight depending on its importance with respect to the target attribute. Besides, our nearest neighbor algorithm gives different weights to the selected instances. Our system is implemented and tested on some typical machine learning databases. Our experiments show that the classification accuracy of our system is, in general, superior to that of other learning methods.<<ETX>>","PeriodicalId":297396,"journal":{"name":"Proceedings of the Tenth Conference on Artificial Intelligence for Applications","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134258935","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 : 1994-03-01DOI: 10.1109/CAIA.1994.323654
H. Hirsh, M. Noordewier
Successful inductive learning requires that training data be expressed in a form where underlying regularities can be recognized by the learning system. Unfortunately, many applications of inductive learning/spl minus/especially in the domain of molecular biology/spl minus/have assumed that data are provided in a form already suitable for learning, whether or not such an assumption is actually justified. This paper describes the use of background knowledge of molecular biology to re-express data into a form more appropriate for learning. Our results show dramatic improvements in classification accuracy for two very different classes of DNA sequences using traditional "off-the-sheIf" decision-tree and neural-network inductive-learning methods.<>
{"title":"Using background knowledge to improve inductive learning of DNA sequences","authors":"H. Hirsh, M. Noordewier","doi":"10.1109/CAIA.1994.323654","DOIUrl":"https://doi.org/10.1109/CAIA.1994.323654","url":null,"abstract":"Successful inductive learning requires that training data be expressed in a form where underlying regularities can be recognized by the learning system. Unfortunately, many applications of inductive learning/spl minus/especially in the domain of molecular biology/spl minus/have assumed that data are provided in a form already suitable for learning, whether or not such an assumption is actually justified. This paper describes the use of background knowledge of molecular biology to re-express data into a form more appropriate for learning. Our results show dramatic improvements in classification accuracy for two very different classes of DNA sequences using traditional \"off-the-sheIf\" decision-tree and neural-network inductive-learning methods.<<ETX>>","PeriodicalId":297396,"journal":{"name":"Proceedings of the Tenth Conference on Artificial Intelligence for Applications","volume":"41 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129830878","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 : 1994-03-01DOI: 10.1109/CAIA.1994.323673
Minhwa Chung, D. Moldovan
Presents a parallel memory-based parser called PARALLEL, which is implemented on a marker-passing parallel AI computer called the Semantic Network Array Processor (SNAP). In the PARALLEL memory-based parser, the parallelism in natural language processing is utilized by a memory search model of parsing. Linguistic information is stored as phrasal patterns in a semantic network knowledge base that is distributed over the memory of the parallel computer. Parsing is performed by recognizing and linking linguistic patterns that reflect a sentence interpretation. This is achieved via propagating markers over the distributed network. We have developed a system capable of processing newswire articles about terrorism with a large knowledge base of 12,000 semantic network nodes. This paper presents the structure of the system, the memory-based parsing method used and performance results obtained.<>
{"title":"Memory-based parsing with parallel marker-passing","authors":"Minhwa Chung, D. Moldovan","doi":"10.1109/CAIA.1994.323673","DOIUrl":"https://doi.org/10.1109/CAIA.1994.323673","url":null,"abstract":"Presents a parallel memory-based parser called PARALLEL, which is implemented on a marker-passing parallel AI computer called the Semantic Network Array Processor (SNAP). In the PARALLEL memory-based parser, the parallelism in natural language processing is utilized by a memory search model of parsing. Linguistic information is stored as phrasal patterns in a semantic network knowledge base that is distributed over the memory of the parallel computer. Parsing is performed by recognizing and linking linguistic patterns that reflect a sentence interpretation. This is achieved via propagating markers over the distributed network. We have developed a system capable of processing newswire articles about terrorism with a large knowledge base of 12,000 semantic network nodes. This paper presents the structure of the system, the memory-based parsing method used and performance results obtained.<<ETX>>","PeriodicalId":297396,"journal":{"name":"Proceedings of the Tenth Conference on Artificial Intelligence for Applications","volume":"196 3","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114044889","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 : 1994-03-01DOI: 10.1109/CAIA.1994.323624
Mostafa Mahmoud Syiam
Presents a neural network expert system to assist a GP in early medical diagnosis of eye diseases in patients. The developed system bases its diagnosis on patient symptoms and signs, and uses a multilayer feedforward network with a single hidden layer. The backpropagation algorithm is employed for training the network in a supervised mode. The effect of the number of nodes in the hidden layer on the developed system's performance is discussed. Analysis of the results indicates that the developed system has a disease diagnosis ratio of above 87 percent. To evaluate the performance of the developed system, a test data set was given to both GPs and specialists. It is indicated that the performance of the developed system exceeds that of the GPs, and it reaches the level of performance of the eye specialists.<>
{"title":"A neural network expert system for diagnosing eye diseases","authors":"Mostafa Mahmoud Syiam","doi":"10.1109/CAIA.1994.323624","DOIUrl":"https://doi.org/10.1109/CAIA.1994.323624","url":null,"abstract":"Presents a neural network expert system to assist a GP in early medical diagnosis of eye diseases in patients. The developed system bases its diagnosis on patient symptoms and signs, and uses a multilayer feedforward network with a single hidden layer. The backpropagation algorithm is employed for training the network in a supervised mode. The effect of the number of nodes in the hidden layer on the developed system's performance is discussed. Analysis of the results indicates that the developed system has a disease diagnosis ratio of above 87 percent. To evaluate the performance of the developed system, a test data set was given to both GPs and specialists. It is indicated that the performance of the developed system exceeds that of the GPs, and it reaches the level of performance of the eye specialists.<<ETX>>","PeriodicalId":297396,"journal":{"name":"Proceedings of the Tenth Conference on Artificial Intelligence for Applications","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123839564","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 : 1994-03-01DOI: 10.1109/CAIA.1994.323628
G. S. Novak
Describes a system that constructs a computer program from a graphical specification provided by the user. The specification consists of diagrams that represent physical and mathematical models; connections between diagram ports signify that corresponding quantities must be equal. A program (in Lisp or C) is generated from the graphical specification by data flow analysis and algebraic manipulation of equations associated with the physical models. Equations, algebraic manipulations, and unit conversions are hidden from the user and are performed automatically. This system allows more rapid generation of programs than would be possible with hand coding.<>
{"title":"Generating programs from connections of physical models","authors":"G. S. Novak","doi":"10.1109/CAIA.1994.323628","DOIUrl":"https://doi.org/10.1109/CAIA.1994.323628","url":null,"abstract":"Describes a system that constructs a computer program from a graphical specification provided by the user. The specification consists of diagrams that represent physical and mathematical models; connections between diagram ports signify that corresponding quantities must be equal. A program (in Lisp or C) is generated from the graphical specification by data flow analysis and algebraic manipulation of equations associated with the physical models. Equations, algebraic manipulations, and unit conversions are hidden from the user and are performed automatically. This system allows more rapid generation of programs than would be possible with hand coding.<<ETX>>","PeriodicalId":297396,"journal":{"name":"Proceedings of the Tenth Conference on Artificial Intelligence for Applications","volume":"41 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124679880","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 : 1994-03-01DOI: 10.1109/CAIA.1994.323634
A. Griffith, R. Simpson, L. Blatt
Studies have shown that software developers tend to use existing human-computer interfaces as examples while designing new interfaces. However, if the examples are poorly designed, or the tasks in the example are inconsistent with the tasks of the new interface, then using examples can be detrimental to the design of the interface. To alleviate the problem of using examples inappropriately, and to support good interface design practices, we are developing the concept of a case-based interface design assistant, called Interface Lab. Interface Lab is a design environment which uses user-centered design, an interface design methodology, as the context for retrieval of cases of interface examples.<>
{"title":"Interface Lab: a case-based interface design assistant","authors":"A. Griffith, R. Simpson, L. Blatt","doi":"10.1109/CAIA.1994.323634","DOIUrl":"https://doi.org/10.1109/CAIA.1994.323634","url":null,"abstract":"Studies have shown that software developers tend to use existing human-computer interfaces as examples while designing new interfaces. However, if the examples are poorly designed, or the tasks in the example are inconsistent with the tasks of the new interface, then using examples can be detrimental to the design of the interface. To alleviate the problem of using examples inappropriately, and to support good interface design practices, we are developing the concept of a case-based interface design assistant, called Interface Lab. Interface Lab is a design environment which uses user-centered design, an interface design methodology, as the context for retrieval of cases of interface examples.<<ETX>>","PeriodicalId":297396,"journal":{"name":"Proceedings of the Tenth Conference on Artificial Intelligence for Applications","volume":"68 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129917599","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 : 1994-03-01DOI: 10.1109/CAIA.1994.323674
H. Almuallim, Y. Akiba, T. Yamazaki, A. Yokoo, S. Kaneda
Addresses the problem of constructing translation rules for ALT-J/E/spl minus/a knowledge-based Japanese-English translation system developed at NTT. We introduce the system ATRACT, which is a semi-automatic knowledge acquisition tool designed to facilitate the construction of the desired translation rules through the use of inductive machine learning techniques. Rather than building rules by hand from scratch, a user of ATRACT can obtain good candidate rules by providing the system with a collection of examples of Japanese sentences along with their English translations. This learning task is characterized by two factors: (i) it involves exploiting a huge amount of semantic information as background knowledge; (ii) training examples are "ambiguous". Currently, two learning methods are available in ATRACT. Experiments show that these methods lead to rules that are very close to those composed manually by human experts given only a reasonable number of examples. These results suggest that ATRACT will significantly contribute to reducing the cost and improving the quality of ALT-J/E translation rules.<>
{"title":"A tool for the acquisition of Japanese-English machine translation rules using inductive learning techniques","authors":"H. Almuallim, Y. Akiba, T. Yamazaki, A. Yokoo, S. Kaneda","doi":"10.1109/CAIA.1994.323674","DOIUrl":"https://doi.org/10.1109/CAIA.1994.323674","url":null,"abstract":"Addresses the problem of constructing translation rules for ALT-J/E/spl minus/a knowledge-based Japanese-English translation system developed at NTT. We introduce the system ATRACT, which is a semi-automatic knowledge acquisition tool designed to facilitate the construction of the desired translation rules through the use of inductive machine learning techniques. Rather than building rules by hand from scratch, a user of ATRACT can obtain good candidate rules by providing the system with a collection of examples of Japanese sentences along with their English translations. This learning task is characterized by two factors: (i) it involves exploiting a huge amount of semantic information as background knowledge; (ii) training examples are \"ambiguous\". Currently, two learning methods are available in ATRACT. Experiments show that these methods lead to rules that are very close to those composed manually by human experts given only a reasonable number of examples. These results suggest that ATRACT will significantly contribute to reducing the cost and improving the quality of ALT-J/E translation rules.<<ETX>>","PeriodicalId":297396,"journal":{"name":"Proceedings of the Tenth Conference on Artificial Intelligence for Applications","volume":"9 1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128896277","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 : 1994-03-01DOI: 10.1109/CAIA.1994.323643
M. Juric
Multiple fault diagnosis (MFD) is the process of determining the correct fault or faults that are responsible for a given set of symptoms. Exhaustive searches or statistical analyses are usually too computationally expensive to solve these types of problems in real-time. We use a simple genetic algorithm to significantly reduce the time required to evolve a satisfactory solution. We show that when using genetic algorithms to solve these kinds of applications, best results are achieved with higher than "normal" mutation rates. Schemata theory is used to analyze this data and show that even though schema length increases, the Hamming distance between binary representations of best-fit chromosomes is quite small. Hamming distance is then related to schema length to show why mutation rate becomes important in this type of application.<>
{"title":"Optimizing genetic algorithm parameters for multiple fault diagnosis applications","authors":"M. Juric","doi":"10.1109/CAIA.1994.323643","DOIUrl":"https://doi.org/10.1109/CAIA.1994.323643","url":null,"abstract":"Multiple fault diagnosis (MFD) is the process of determining the correct fault or faults that are responsible for a given set of symptoms. Exhaustive searches or statistical analyses are usually too computationally expensive to solve these types of problems in real-time. We use a simple genetic algorithm to significantly reduce the time required to evolve a satisfactory solution. We show that when using genetic algorithms to solve these kinds of applications, best results are achieved with higher than \"normal\" mutation rates. Schemata theory is used to analyze this data and show that even though schema length increases, the Hamming distance between binary representations of best-fit chromosomes is quite small. Hamming distance is then related to schema length to show why mutation rate becomes important in this type of application.<<ETX>>","PeriodicalId":297396,"journal":{"name":"Proceedings of the Tenth Conference on Artificial Intelligence for Applications","volume":"121 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116916694","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 : 1994-03-01DOI: 10.1109/CAIA.1994.323659
G. Biswas, G. Lee
Discusses the application of conceptual clustering in restructuring large knowledge bases for the purpose of improving their complex problem solving efficiency. The rule base of PLAYMAKER, a system for characterizing hydrocarbon fields and plays, is restructured into a hierarchy of rule models using our conceptual clustering scheme, ITERATE. The rule models, used with a task-specific reasoning methodology, provide a more efficient, focused, and robust inferencing mechanism. A set of case studies that have been conducted demonstrate the improved performance of the reasoning system. PLAYMAKER is implemented on MIDST (Mixed Inferencing Dempster-Shafer Tool), a general-purpose knowledge-based system construction tool that incorporates reasoning mechanisms based on a task-specific architecture and belief functions.<>
{"title":"Knowledge reorganization. A rule model scheme for efficient reasoning","authors":"G. Biswas, G. Lee","doi":"10.1109/CAIA.1994.323659","DOIUrl":"https://doi.org/10.1109/CAIA.1994.323659","url":null,"abstract":"Discusses the application of conceptual clustering in restructuring large knowledge bases for the purpose of improving their complex problem solving efficiency. The rule base of PLAYMAKER, a system for characterizing hydrocarbon fields and plays, is restructured into a hierarchy of rule models using our conceptual clustering scheme, ITERATE. The rule models, used with a task-specific reasoning methodology, provide a more efficient, focused, and robust inferencing mechanism. A set of case studies that have been conducted demonstrate the improved performance of the reasoning system. PLAYMAKER is implemented on MIDST (Mixed Inferencing Dempster-Shafer Tool), a general-purpose knowledge-based system construction tool that incorporates reasoning mechanisms based on a task-specific architecture and belief functions.<<ETX>>","PeriodicalId":297396,"journal":{"name":"Proceedings of the Tenth Conference on Artificial Intelligence for Applications","volume":"102 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124163706","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 : 1994-03-01DOI: 10.1109/CAIA.1994.323633
X. Guan, R. Mural, E. Uberbacher
Describes a new approach for predicting protein structures based on artificial intelligence methods and genetic algorithms. We combine nearest neighbor searching algorithms, neural networks, heuristic rules and genetic algorithms to form an integrated system to predict protein structures from their primary amino acid sequences. First, we describe our methods and how they are integrated, and then apply our methods to several protein sequences. The results are very close to the real structures obtained by crystallography. Parallel genetic algorithms are also implemented.<>
{"title":"Protein structure prediction using hybrid AI methods","authors":"X. Guan, R. Mural, E. Uberbacher","doi":"10.1109/CAIA.1994.323633","DOIUrl":"https://doi.org/10.1109/CAIA.1994.323633","url":null,"abstract":"Describes a new approach for predicting protein structures based on artificial intelligence methods and genetic algorithms. We combine nearest neighbor searching algorithms, neural networks, heuristic rules and genetic algorithms to form an integrated system to predict protein structures from their primary amino acid sequences. First, we describe our methods and how they are integrated, and then apply our methods to several protein sequences. The results are very close to the real structures obtained by crystallography. Parallel genetic algorithms are also implemented.<<ETX>>","PeriodicalId":297396,"journal":{"name":"Proceedings of the Tenth Conference on Artificial Intelligence for Applications","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125719463","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：