In this paper, we consider the task of automatic synthesis/leaming of patternrecognition systems. In particular, a method is proposed that, given exclusively training raster images, synthesizes complete feature-based...
详细信息
ISBN:
(纸本)3540405046
In this paper, we consider the task of automatic synthesis/leaming of patternrecognition systems. In particular, a method is proposed that, given exclusively training raster images, synthesizes complete feature-based recognition system. The proposed approach is general and does not require any assumptions concerning training data and application domain. Its novelty consists in procedural representation of features for recognition and utilization of coevolutionary computation for their synthesis. The paper describes the synthesis algorithm, outlines the architecture of the synthesized system, provides firm rationale for its design, and evaluates it experimentally on the real-world task of target recognition in synthetic aperture radar (SAR) imagery.
Case Based Reasoning systems rely on competent case knowledge for effective problem-solving. However, for many problem solving tasks, notably design, simple retrieval from the case-base in not sufficient. Further know...
详细信息
ISBN:
(纸本)3540405046
Case Based Reasoning systems rely on competent case knowledge for effective problem-solving. However, for many problem solving tasks, notably design, simple retrieval from the case-base in not sufficient. Further knowledge is required to help effective retrieval and to undertake adaptation of the retrieved solution to suit the new problem better. This paper proposes methods to learn knowledge for the retrieval and adaptation knowledge containers exploiting the knowledge already captured in the case knowledge.
The Dirichlet distribution offers high flexibility for modeling data. This paper describes two new mixtures based on this density: the GDD (Generalized Dirichlet Distribution) and the MDD (Multinomial Dirichlet Distri...
详细信息
ISBN:
(纸本)3540405046
The Dirichlet distribution offers high flexibility for modeling data. This paper describes two new mixtures based on this density: the GDD (Generalized Dirichlet Distribution) and the MDD (Multinomial Dirichlet Distribution) mixtures. These mixtures will be used to model continuous and discrete data, respectively. We propose a method for estimating the parameters of these mixtures. The performance of our method is tested by contextual evaluations. In these evaluations we compare the performance of Gaussian and GDD mixtures in the classification of several pattern-recognitiondata sets and we apply the MDD mixture to the problem of summarizing image databases.
This paper handles the integration of fuzziness with On-Line Analytical Processing (OLAP)(1) association rules mining, It contributes to the ongoing research on multidimensional online datamining by proposing a gener...
详细信息
ISBN:
(纸本)3540405046
This paper handles the integration of fuzziness with On-Line Analytical Processing (OLAP)(1) association rules mining, It contributes to the ongoing research on multidimensional online datamining by proposing a general architecture that uses a fuzzy data cube for knowledge discovery. Three different methods are introduced to mine fuzzy association rules in the constructed fuzzy data cube, namely single dimension, multidimensional and hybrid association rules mining;the third structure integrates the other two methods. To the best of our knowledge, this is the first effort in this direction. Experimental results obtained for each of the three methods on the adult data of the United States census in 2000 show the effectiveness and applicability of the proposed mining approach.
Multi-label decision procedures are the target of the supervised learning algorithm we propose in this paper. Multi-label decision procedures map examples to a finite set of labels. Our learning algorithm extends Scha...
详细信息
ISBN:
(纸本)3540405046
Multi-label decision procedures are the target of the supervised learning algorithm we propose in this paper. Multi-label decision procedures map examples to a finite set of labels. Our learning algorithm extends Schapire and Singer's *** and produces sets of rules that can be viewed as trees like Alternating Decision Trees (invented by Freund and Mason). Experiments show that we take advantage of both performance and readability using boosting techniques as well as tree representations of large set of rules. Moreover, a key feature of our algorithm is the ability to handle heterogenous input data: discrete and continuous values and text data.
作者:
Armengol, EPlaza, ECSIC
Spanish Council Sci Res Artificial Intelligence Res Inst Catalonia 08193 Spain
In concept learning, inductive techniques perform a global approximation to the target concept. Instead, lazy learning techniques use local approximations to form an implicit global approximation of the target concept...
详细信息
ISBN:
(纸本)3540405046
In concept learning, inductive techniques perform a global approximation to the target concept. Instead, lazy learning techniques use local approximations to form an implicit global approximation of the target concept. In this paper we present C-LID, a lazy learning technique that uses LID for generating local approximations to the target concept. LID generates local approximations in the form of similitude terms (symbolic descriptions of what is shared by 2 or more cases). C-LID caches and reuses the similitude terms generated in past cases to improve the problem solving of future problems. The outcome of C-LID (and LID) is assessed with experiments on the Toxicology dataset.
Most recent document standards rely on structured representations. On the other hand, current information retrieval systems have been developed for flat document representations and cannot be easily extended to cope w...
详细信息
ISBN:
(纸本)3540405046
Most recent document standards rely on structured representations. On the other hand, current information retrieval systems have been developed for flat document representations and cannot be easily extended to cope with more complex document types. Only a few models have been proposed for handling structured documents, and the design of such systems is still an open problem. We present here a new model for structured document retrieval which allows to compute and to combine the scores of document parts. It is based on bayesian networks and allows for learning the model parameters in the presence of incomplete data. We present an application of this model for ad-hoc retrieval and evaluate its performances on a small structured collection. The model can also be extended to cope with other tasks such as interactive navigation in structured documents or corpus.
When modeling technical processes, the training data regularly come from test plans, to reduce the number of experiments and to save time and costs. On the other hand, this leads to unobserved combinations of the inpu...
详细信息
ISBN:
(纸本)3540405046
When modeling technical processes, the training data regularly come from test plans, to reduce the number of experiments and to save time and costs. On the other hand, this leads to unobserved combinations of the input variables. In this article it is shown, that these unobserved configurations might lead to un-trainable parameters. Afterwards a possible design criterion is introduced, which avoids this drawback. Our approach is tested to model a welding process. The results show, that hybrid Bayesian networks are able to deal with yet unobserved in- and output data.
Automatically authoring or acquiring cases in the case-based reasoning (CBR) systems is recognized as a bottleneck issue that can determine whether a CBR system will be successful or not. In order to reduce human effo...
详细信息
ISBN:
(纸本)3540405046
Automatically authoring or acquiring cases in the case-based reasoning (CBR) systems is recognized as a bottleneck issue that can determine whether a CBR system will be successful or not. In order to reduce human effort required for authoring the cases, we propose a framework for authoring the case from the unstructured, free-text, historic maintenance data by applying natural language processing technology. This paper provides an overview of the proposed framework, and outlines its implementation, an automated case creation system for the Integrated Diagnostic System. Some experimental results for testing the framework are also presented.
With the electricity market liberalization, the distribution and retail companies are looking for better market strategies based on adequate information upon the consumption patterns of its electricity customers. A fa...
详细信息
ISBN:
(纸本)3540405046
With the electricity market liberalization, the distribution and retail companies are looking for better market strategies based on adequate information upon the consumption patterns of its electricity customers. A fair insight on the customers' behavior will permit the definition of specific contract aspects based on the different consumption patterns. In this paper, we propose a KDD project applied to electricity consumption data from a utility client's database. To form the different customers' classes, and find a set of representative consumption patterns, a comparative analysis of the performance of the K-means, Kohonen Self-Organized Maps (SOM) and a Two-Level approach is made. Each customer class will be represented by its load profile obtained with the algorithm with best performance in the data set used.
暂无评论