Fuzzy Rule-Based System (FRB) in the form of human comprehensible IF-THEN rules can be extracted from Support Vector machine (SVM) which is regarded as a black-boxed system. We first prove that SVM decision network an...
详细信息
ISBN:
(纸本)9783642200427
Fuzzy Rule-Based System (FRB) in the form of human comprehensible IF-THEN rules can be extracted from Support Vector machine (SVM) which is regarded as a black-boxed system. We first prove that SVM decision network and the zero-ordered Sugeno FRB type of the Adaptive Network Fuzzy Inference System (ANFIS) are equivalent indicating that SVM's decision can actually be represented by fuzzy IF-THEN rules. We then propose a rule extraction method based on kernel function firing strength and unbounded support vector space expansion. An advantage of our method is the guarantee that the number of final fuzzy IF-THEN rules is equal or less than the number of support vectors in SVM, and it may reveal human comprehensible patterns. We compare our method against SVM using popular benchmark data sets, and the results are comparable.
In this paper, we explain the bag of words representation from a soft computing perspective. The traditional Bag of word representation describes an image as a bag of discrete visual codewords. Where histogram of the ...
详细信息
In this study an approach for detecting biomedical syntax variations through the Named Entity recognition (NER) called Statistical Character-Based Syntax Similarity (SCSS) is proposed which is used by dictionary-based...
详细信息
ISBN:
(纸本)9783642221842
In this study an approach for detecting biomedical syntax variations through the Named Entity recognition (NER) called Statistical Character-Based Syntax Similarity (SCSS) is proposed which is used by dictionary-based NER approaches. Named Entity recognition for biomedical literatures is extraction and recognition of biomedical names. There are different types of NER approaches, that the most common one is dictionary-based approaches. For a given unknown pattern, Dictionary-Based approaches, search through a biomedical dictionary and finds the most common similar patterns to assign their biomedical types to the given unknown pattern. Biomedical literatures include syntax variations, which means two different patterns, refer to the same biomedical named entity. Hence a similarity function should be able to support all of the possible syntax variations. There are three syntax variations namely: (i) character-level, (ii) word-level, and (iii) word order. The SCSS is able to detect all of the mentioned syntax vitiations. This study is evaluated based on two measures: recall and precision which are used to calculate a balanced F-score. Result is satisfied as recall is 92.47% and precision is 96.7%, while the f-test is 94.53%.
Off-line signature verification is an important form of behavioral biometric identification. We present;a method utilizing Modified Direction Feature(MDF) and Microstructure Feature(MSF) to tackle the problem. MDF and...
详细信息
ISBN:
(纸本)9783642271823
Off-line signature verification is an important form of behavioral biometric identification. We present;a method utilizing Modified Direction Feature(MDF) and Microstructure Feature(MSF) to tackle the problem. MDF and MSF belong to geometric structure features, but these two features are different from each other in each emphasis. In our study, global information in signatures' boundaries is represented by MDF, while local information is represented by MSF. In order to get features with lower dimensions, principal component analysis is employed to reduce redundant dimensions. In addition, we adopt support vector machine as classifier for verification process. The proposed strategy is evaluated on the GPDS and MCY'T corpora. Experimental results have demonstrated that;the proposed method is effective to improve off-line signature verification accuracy.
Magnetic and inductive sensors are widely used in research and industry for a variety of applications, e. g. for geophysical prospecting, non-destructive testing (NDT) of materials, in the food industries, for distanc...
详细信息
ISBN:
(纸本)9783642215377
Magnetic and inductive sensors are widely used in research and industry for a variety of applications, e. g. for geophysical prospecting, non-destructive testing (NDT) of materials, in the food industries, for distance and proximity sensing, security systems as well as for landmine detection. Metal detectors (MD), based on the eddy current principle, are the most used systems in humanitarian demining. Their main disadvantage is the high false alarm rate, caused by harmless metal objects and "uncooperative" soils with magnetic properties. The sensor signal (i.e. the induced complex coil voltage) is influenced by the object properties (material, shape). This paper describes an object recognition based on multi-parameter MD signals, which are classified by the fuzzy method. The ability to identify mines by their characteristic signature was demonstrated in test lanes for mine detection provided by the University of Rostock (Germany), JRC-Ispra (Italy) and the CTRO-Benkovac (Croatia).
Texture analysis is used in numerous applications in various fields. There have been many different approaches/techniques in the literature for texture analysis among which the texton-based approach that computes the ...
详细信息
ISBN:
(纸本)9783642215926;9783642215933
Texture analysis is used in numerous applications in various fields. There have been many different approaches/techniques in the literature for texture analysis among which the texton-based approach that computes the primitive elements representing textures using k-means algorithm has shown great success. Recently, dictionary learning and sparse coding has provided state-of-the-art results in various applications. With recent advances in computing the dictionary and sparse coefficients using fast algorithms, it is possible to use these techniques to learn the primitive elements and histogram of them to represent textures. In this paper, online learning is used as fast implementation of sparse coding for texture classification. The results show similar to or better performance than texton based approach on CUReT database despite of computation of dictionary without taking into account the class labels.
In this paper, we propose a new approach for detecting people in video sequences based on geometrical features and AdaBoost learning. Unlike its predecessors, our approach uses features calculated directly from silhou...
详细信息
ISBN:
(纸本)9783642215926;9783642215933
In this paper, we propose a new approach for detecting people in video sequences based on geometrical features and AdaBoost learning. Unlike its predecessors, our approach uses features calculated directly from silhouettes produced by change detection algorithms. Moreover, feature analysis is done part by part for each silhouette, making our approach efficiently applicable for partially-occluded pedestrians and groups of people detection. Experiments on real-world videos showed us the performance of the proposed approach for real-time pedestrian detection.
This paper presents an iterative Content Based Image Retrival(CBIR) system with Relevance Feedback (RF), in which M-band wavelet features are used as representation of images. The pixels are clustered using Fuzzy C-Me...
详细信息
ISBN:
(纸本)9783642217869
This paper presents an iterative Content Based Image Retrival(CBIR) system with Relevance Feedback (RF), in which M-band wavelet features are used as representation of images. The pixels are clustered using Fuzzy C-Means (FCM) clustering algorithm to obtain an image signature and Earth Mover's Distance (EMD) is used as a distance measure. Fuzzy entropy based feature evaluation mechanism is used for automatic computation of revised feature importance and similarity distance at the end of each iteration. The performance of the algorithm is tested on standard large multi-class image databases and compared with MPEG-7 visual features.
This article discusses a new document indexing scheme for information retrieval. For a structured (e.g., scientific) document, Pasi et al. proposed varying weights to different sections according to their importance i...
详细信息
ISBN:
(纸本)9783642217869
This article discusses a new document indexing scheme for information retrieval. For a structured (e.g., scientific) document, Pasi et al. proposed varying weights to different sections according to their importance in the document. This concept is extended here to unstructured documents. Each sentence in a document is initially assigned weight (significance in the document) with the help of a summarization technique. Accordingly, the term frequency of a term is decided as the sum of weights of the sentences the term belongs. The method is verified on a real life dataset using leading existing information retrieval models, and its performance has been found to be superior to conventional indexing schemes.
The spermatogenesis is crucial to the species reproduction, and its monitoring may shed light over some important information of such process. Thus, the germ cells quantification can provide useful tools to improve th...
详细信息
ISBN:
(纸本)9781424441228
The spermatogenesis is crucial to the species reproduction, and its monitoring may shed light over some important information of such process. Thus, the germ cells quantification can provide useful tools to improve the reproduction cycle. In this paper, we present the first work that address this problem in fishes with machine learning techniques. We show here how to obtain high recognition accuracies in order to identify fish germ cells with several state-of-the-art supervised patternrecognition techniques.
暂无评论