The Infona portal uses cookies, i.e. strings of text saved by a browser on the user's device. The portal can access those files and use them to remember the user's data, such as their chosen settings (screen view, interface language, etc.), or their login data. By using the Infona portal the user accepts automatic saving and using this information for portal operation purposes. More information on the subject can be found in the Privacy Policy and Terms of Service. By closing this window the user confirms that they have read the information on cookie usage, and they accept the privacy policy and the way cookies are used by the portal. You can change the cookie settings in your browser.
In multi-class problems, within- and between-class scatters should be considered in classification criterion. The common vector approach (CVA) uses the discriminative information obtained from within-class scatter of any class. It has been shown that this classical CVA method gives high recognition rates in multi-class problems. In this study, improvements on the CVA method that consider both within-...
Voice conversion techniques enable the transformation of a source speaker's voice to that of a target speaker's automatically. The performance of any voice conversion algorithm depends on the source-target pair chosen. This study focuses on the problem of source speaker (donor) selection from a set of available speakers that will result in the best quality output for a specific target speaker's voice...
We present a method for combining a number of Support Vector Machines trained independently in the eigenface space and we apply it to face class modeling. We first train several SVMs on subsets of some initial training set and then combine their expertise using various probabilistic combining rules. This approach is compared to a classical SVM classification as well as Multiple SVM classification[1].
In this paper, we investigate the use of discriminant feature selection techniques in the elastic graph matching (EGM) algorithm. State of the art and novel discriminant dimensionality reduction techniques are used in the node feature vectors in order to extract discriminant features. We illustrate the improvements in performance in frontal face verification using a modified multiscale morphological...
In this paper, a new approach which is called the common matrix approach is proposed for face recognition. The common matrix for each class can be calculated either using Gram-Schmidt orthogonalization method or using scatter matrix of each class. In both ways, orthonormal mat rices in the indifference subspace represent the directions that contain important discriminative information. The proposed...
This paper presents an automatic speaker recognition system for intelligence applications. The system has to provide functionalities for a speaker skimming application in which databases of recorded conversations belonging to an ongoing investigation can be annotated and quickly browsed by an operator. The paper discusses the criticalities introduced by the characteristics of the audio signals under...
The purpose of this study is to investigate the effects of different forms of between-class scatter matrices on multi-class problems. Two different between-class scatter matrices are defined in Fisher's linear discriminant analysis (FLDA) and the classification rates better than that of classical FLDA are obtained for TI-digit database. In this study, the criteria that give separate subspaces for...
In this paper we present our recent work in implementing Serbian spoken dialogue system for the bus information retrieval at the main Belgrade bus station. Dialogue is organized into several levels. At each level, system has to recognize a limited number of keywords in continuous speech of Serbian. The keywords were modeled by HMMs (Hidden Markov Models) in such a way that each syllable is three-state...
We present a method to improve the performance of content-based image retrieval (CBIR) systems. The idea is based on the concept of query models [1], which generalizes the notion of similarity in multi-feature queries. In a query model features are organized in layers. Each succeeding layer has to investigate only a subset of the image set the preceding layer had to examine. For the purpose of performance...
Score normalisation with cohort speaker models has been widely used in HMM-based speaker verification. Most of the proposed methods are based on the framework of the hypothesis testing. Based on this framework an overall average of all cohort scores is often used for normalisation, which leads a log likelihood ratio (LLR) for verification. In this paper we use a competition-based criterion to define...
This paper applies pyramidal multiresolution design, which has been used to design binary filters, to design aperture filters, which are nonlinear gray-scale filters that operate on data in a range-domain window (aperture). The designed filters are used to find markers for the region of the eyes in an image database of faces.
The concept of optimal hyperplane has been recently proposed in the context of statistical learning theory. The important property of an optimal hyperplane is that it provides maximum margins to each class to be separated. Obviously, such a decision boundary is expected to yield good generalization. Currently, the support vector machines (SVM) are probably one of the very few models (if not the only...
In this paper, we propose a general purpose video-based tracking system that is robust to occlusion. The system is tracking moving objects that are represented by connected regions of pixels that are tracked over the video frames. The proposed tracking system design is based on recent ideas and results on robust set-valued state estimation for uncertain control systems.
Short-term load forecasting is the basic work for each power system economic dispatch, also, optimal combination unit and optimal power flow is the prerequisite for real-time electricity market trading. This paper presents a new evaluated point weighted method, which is based on distance vector clustering analyzes maximum load and holidays influence. Actual system application results show that the...
In this study, a new method for the detection of T wave alternans in multichannel ECG signals is introduced. The use of tensors (multidimensional matrices) allows us to combine the information present in all channels, making detection more robust. To construct a 3D tensor from a 2D ECG signal, the T wave is first roughly segmented. The intervals are then placed after each other to obtain a 3D structure...
Conventional speaker identification systems are already field-proven with respect to recognition accuracy. Since any biometric identification requires exhaustive 1 : N comparisons for identifying a biometric probe, comparison time frequently dominates the overall computational workload, preventing the system from being executed in real-time. In this paper we propose a computational efficient two-stage...
Common image features have too poor information for identification of forensic images of fingerprints, where only a small area of the finger is imaged and hence a small amount of key points are available. Noise, nonlinear deformation, and unknown rotation are additional issues that complicate identification of forensic fingerprints. We propose a feature extraction method which describes image information...
Grid systems have emerged as a means of sharing computational resources and information. Providing services for accessing, sharing and modifying large databases is a crucial task for grid management systems. This paper proposes an artificial neural network (ANN) prediction mechanism that provides an enhancement to data replication solutions within grid systems. Current replication services often exhibit...
We propose a new ECG data compression algorithm based on a learned overcomplete dictionary to exploit the correlation between signals in adjacent heart beats. The learned overcomplete dictionary is constructed by K-SVD dictionary learning algorithm, after preprocessing and normalization of length and magnitude. Using the overcomplete dictionary, the proposed algorithm can find sparse estimation, which...
Injections flaws which include SQL injection are the most prevalent security threats affecting Web applications[1]. To mitigate these attacks, Web Application Firewalls (WAFs) apply security rules in order to both inspect HTTP data streams and detect malicious HTTP transactions. Nevertheless, attackers can bypass WAF's rules by using sophisticated SQL injection techniques. In this paper, we introduce...
Set the date range to filter the displayed results. You can set a starting date, ending date or both. You can enter the dates manually or choose them from the calendar.