This paper proposes a new corrective learning algorithm and evaluates the performance by handwritten numeral recognition test. The algorithm generates a mirror image of a pattern which belongs to one class of a pair o...
详细信息
ISBN:
(纸本)3540423591
This paper proposes a new corrective learning algorithm and evaluates the performance by handwritten numeral recognition test. The algorithm generates a mirror image of a pattern which belongs to one class of a pair of confusing classes and utilizes it as a learningpattern of the other class. statistical patternrecognition techniques generally assume that the density function and the parameters of each class are only dependent on the sample of the class. The mirror image learning algorithm enlarges the learning sample of each class by mirror image patterns of other classes and enables us to achieve higher recognition accuracy with small learning sample. recognition accuracies of the minimum distance classifier and the projection distance method were improved from 93.17% to 98.38% and from 99.11% to 99.37% respectively in the recognition test for handwritten numeral database IPTP CD-ROM1 [1].
In recent years feedback approaches have been used in relating low-level image features with concepts to overcome the subjective nature of the human image interpretation. Generally, in these systems when the user star...
详细信息
ISBN:
(纸本)3540423591
In recent years feedback approaches have been used in relating low-level image features with concepts to overcome the subjective nature of the human image interpretation. Generally, in these systems when the user starts with a new query, the entire prior experience of the system is lost. In this paper, we address the problem of incorporating prior experience of the retrieval system to improve the performance on future queries. We propose a semi-supervised fuzzy clustering method to learn class distribution (meta knowledge) in the sense of high-level concepts from retrieval experience. Using fuzzy rules, we incorporate the meta knowledge into a probabilistic relevance feedback approach to improve the retrieval performance. Results presented on synthetic and real databases show that our approach provides better retrieval precision compared to the case when no retrieval experience is used.
How to increase both autonomy and versatility of a knowledge discovery system is a core problem and a crucial aspect of KDD (Knowledge Discovery and datamining). Within the framework of the KDD process and the GLS (G...
详细信息
How to increase both autonomy and versatility of a knowledge discovery system is a core problem and a crucial aspect of KDD (Knowledge Discovery and datamining). Within the framework of the KDD process and the GLS (Global learning Scheme) system recently proposed by us, this paper describes a way of increasing both autonomy and versatility of a KDD system by dynamically organizing KDD processes. In our approach, the KDD process is modeled as an organized society of KDD agents with multiple levels. We propose an ontology to describe KDD agents, in the style of GOER (Object Oriented Entity Relationship) data model. Based on this ontology of KDD agents, we apply several AI planning techniques, which are implemented as a meta-agent, so that we might (1) solve the most difficult problem in a multiagent KDD system: how to automatically choose appropriate KDD techniques (KDD agents) to achieve a particular discovery goal in a particular application domain;(2) tackle the complexity of KDD process;and (3) support evolution of KDD data, knowledge and process. The GLS system, as a multistrategy and multiagent KDD system based on the methodology, increases both autonomy and versatility.
A large amount of information is stored in databases, in intranets or in Internet. This information is organised in documents or in text documents. The difference depends on the fact if pictures, tables, figures, and ...
详细信息
ISBN:
(纸本)3540423591
A large amount of information is stored in databases, in intranets or in Internet. This information is organised in documents or in text documents. The difference depends on the fact if pictures, tables, figures, and formulas are included or not. The common problem is to find the desired piece of information, a trend, or an undiscovered pattern from these sources. The problem is not a new one. Traditionally the problem has been considered under the title of information seeking, this means the science how to find a book in the library. Traditionally the problem has been solved either by classifying and accessing documents by Dewey Decimal Classification system or by giving a number of characteristic keywords. The problem is that nowadays there axe lots of unclassified documents in company databases and in intranet or in Internet. First one defines some terms. Text filtering means an information seeking process in which documents are selected from a dynamic text stream. Text mining is a process of analysing text to extract information from it for particular purposes. Text categorisation means the process of clustering similar documents from a large document set. All these terms have a certain degree of overlapping. Text mining, also know as document information mining, text datamining, or knowledge discovery in textual databases is an merging technology for analysing large collections of unstructured documents for the purposes of extracting interesting and non-trivial patterns or knowledge. Typical subproblems that have been solved axe language identification, feature selection/extraction, clustering, natural language processing, summarisation, categorisation, search, indexing, and visualisation. These subproblems are discussed in detail and the most common approaches axe given. Finally some examples of current uses of text mining are given and some potential application areas are mentioned.
This paper proposes a general local learning framework to effectively alleviate the complexities of classifier design by means of "divide and conquer" principle and ensemble method. The learning framework co...
详细信息
Although neural networks have many appealing properties, yet there is neither a systematic way how to set up the topology of a neural network nor how to determine its various learning parameters. Thus an expert is nee...
详细信息
We describe a fuzzy inference approach for detecting and classifying shot transitions in video sequences. Our approach basically extends FAM(Fuzzy Associative Memory) to detect and classify shot transitions, including...
详细信息
The proceedings contain 38 papers. The special focus in this conference is on Multiple Classifier Systems. The topics include: Ensemble methods in machinelearning;experiments with classifier combining rules;a survey ...
ISBN:
(纸本)3540677046
The proceedings contain 38 papers. The special focus in this conference is on Multiple Classifier Systems. The topics include: Ensemble methods in machinelearning;experiments with classifier combining rules;a survey of sequential combination of word recognizers in handwritten phrase recognition at CEDAR;multiple classifier combination methodologies for different output levels;a mathematically rigorous foundation for supervised learning;implementations and theoretical issues;some results on weakly accurate base learners for boosting regression and classification;complexity of classification problems and comparative advantages of combined classifiers;effectiveness of error correcting output codes in multiclass learning problems;combining fisher linear discriminants for dissimilarity representations;a learning method of feature selection for rough classification;analysis of a fusion method for combining marginal classifiers;a hybrid projection based and radial basis function architecture;combining multiple classifiers in probabilistic neural networks;supervised classifier combination through generalized additive multi-model;dynamic classifier selection;boosting in linear discriminant analysis;different ways of weakening decision trees and their impact on classification accuracy of DT combination;applying boosting to similarity literals for time series classification;boosting of tree-based classifiers for predictive risk modeling in GIS;a new evaluation method for expert combination in multi-expert system designing;diversity between neural networks and decision trees for building multiple classifier systems;self-organizing decomposition of functions;classifier instability and partitioning;a hierarchical multiclassifier system for hyperspectral data analysis and consensus based classification of multisource remote sensing data.
Ensemble methods are learning algorithms that construct a set of classifiers and then classify new data points by taking a (weighted) vote of their predictions. The original ensemble method is Bayesian averaging, but ...
详细信息
ISBN:
(数字)9783540450146
ISBN:
(纸本)3540677046
Ensemble methods are learning algorithms that construct a set of classifiers and then classify new data points by taking a (weighted) vote of their predictions. The original ensemble method is Bayesian averaging, but more recent algorithms include error-correcting output coding, Bagging, and boosting. This paper reviews these methods and explains why ensembles can often perform better than any single classifier. Some previous studies comparing ensemble methods are reviewed, and some new experiments are presented to uncover the reasons that Adaboost does not overfit rapidly.
Investigating a data set of the critical size makes a classification task difficult. studying dissimilarity data refers to such a problem, since the number of samples equals their dimensionality. In such a case, a sim...
详细信息
暂无评论