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.
Sentiment classification is classification of reviews into positive or negative depends on the sentiment words expressed in reviews. Generally, sentiments are expressed differently in different domain and annotating label for every domain is expensive and time consuming. In cross domain sentiment classification, a classifier trained in source domain is applied to classify reviews of target domain...
There are many challenges for sentiment classification of user-generated content (UGC) on social media platforms such as micro-blogs. Context dependence, which has been the most challenging problem, is focused on in this paper, and a novel semi-supervised framework is proposed to address the problem. By dividing the feature space of sentiment classification into two parts including the general features...
In this paper, we investigated Deep Learning (DL) for characterizing and detecting target images in an image rapid serial visual presentation (RSVP) task based on EEG data. We exploited DL technique with input feature clusters to handle high dimensional features related to time - frequency events. The method was applied to EEG recordings of a RSVP experiment with multiple sessions and subjects. For...
Object recognition and categorization are two important key features of computer vision. Accuracy aspects represent research challenge fo r both object recognition and categorization techniques. High performance computing (HPC) technologies usually manage the increasing time and complexity of computations. In this paper, a new approach that use 3D spin-images for 3D object categorization is introduced...
Pedestrian detection technology which is based on vision is one of the core technologies in the intelligent transportation system. This paper focuses on the image and video in traffic system. The image and video are normalized and are extracted of histograms of oriented gradients (HOG) by using the linear support vector machine (SVM) to construct efficient classifier. It can realize accurate detection...
This paper describes a new implementation of a mixture of techniques not used before for fingerprint recognition. The implementation consists of three stages: the location of the core, which is done through Radon transformation, the extraction of features (out of which a square fingerprint is produced with the core, and the center of the mass is obtained from it), in stage three, the resulting image...
Using high-dimensional Joint Factor Analysis (JFA) speaker supervectors for the Fishervoice based subspace analysis suffers high computational complexity problem in the model training process. To address this problem, we propose a two-level sampling subspace framework. For the first level of this framework, partial mean vectors are selected from the JFA speaker supervector to form a low-dimensional...
In this paper, a specific region called affine noisy invariant region is extracted from a query and database images to help accurate retrieval on different attacks. Then, only a 64×1 codebook based feature vector is obtained from this specific region applying vector quantization and codebook generation based on the Linde-Buzo-Gray algorithm, which reduces retrieval feature comparison calculations...
We present an offline signature verification system using three different pseudo-dynamic features, two different classifier training approaches and two datasets. One of the most difficult problems of off-line signature verification is that the signature is just a static image while losing a lot of useful dynamic information. Three separate pseudo-dynamic features based on gray level: local binary...
In this paper, a pattern classification task was regarded as a sample selection problem where a sparse subset of sample from the labeled training set was chosen. We proposed an adaptive learning algorithm utilizing the least square function to address this problem. Using these selected samples, which we call informative vectors, a classifier capable of recognizing the test samples was established...
In order to solve the problems of traditional machine learning methods for automatic classification of vulnerability, this paper presents a novel machine learning method based on LDA model and SVM. Firstly, word location information is introduced into LDA model called WL-LDA (Weighted Location LDA), which could acquire better effect through generating vector space on themes other than on words. Secondly,...
To handle high dimensional variables in real world, especially multimedia data, dimension reduction techniques provide effective solutions for feature selection which makes the problem easy to deal in a lower dimension subspace. However, the primary problem with traditional dimension reduction method is to estimate intrinsic dimension of manifold supporting the raw data. Since all existing approaches...
This paper describes a novel incremental PSVM to incorporate new target class information unavailable previously in the underwater target classification system. It is capable of updating already existing multiclass ‘One against Rest’ Proximal Support Vector Machine classifier on arrival of features of new classes. The performance of the algorithm is studied on real data. Simulation establishes the...
This paper presents a method of modern deep machine learning and its application in dimension reduction and lossy compression. Deep belief networks (or DBN's), first proposed by Yoshua Bengio, are hierarchic, stochastic, neural networks with appropriate architecture and dedicated training algorithms. They are composed of layers, each one of which is a restricted Boltzmann machine (or RBM). There exists...
In this paper, a novel method of target identification in foliage environment is presented. This method takes the received signal waveforms to identify the targets between the communication transceivers, which are measured by Ultra WideBand (UWB) Impulse Radio (IR) equipment under foliage environment. In this way, most existing UWB-IR transceivers can be exploited as detecting radar sensors, which...
Based on the analysis of the Hadoop open source distributed computing platform as well as the parallel training methods for the BP network, for the disadvantage of time-consuming when using large amounts of texts to train the BP network, we designed a BP network text categorization model based on data parallel method on Hadoop platform using the MapReduce programming model. The model uses the method...
k-Nearest Neighbor (KNN) algorithm was an efficient text categorization algorithm in recall and accuracy, but the computational overhead of KNN was directly proportional to the sample size, so its classification speed was low in large-scale sample data. Aiming at this problem, the paper presented a density-based method for reducing training data, the method clustered each class of sample data into...
In order to enhance security and protection capability, the integration of different biometric features to set up multimodal biometric authentication system is an effective way. It can provide complementary information to enhance recognition rate, and it can further enhance the reliability and stability of the identity authentication system. However, although the use of multimodal biometric feature...
This Paper presents an SVM (Support Vector Machine) based real-time pedestrian detection scheme for next-generation automotive vision applications. To meet the requirement of real-time detection with high accuracy, we designed the proposed system consisting of 2-stage hierarchical SVMs. In the proposed system, most of the input data are classified by the 1st stage linear SVM and only the inputs between...
In this paper, we propose a novel model for Document Representation in an attempt to address the problem of huge dimensionality and vector sparseness that are commonly faced in Text Classification tasks. We conduct our experiments on data sets of Opinion Mining. We use as classifiers Support Vector Machines (SVM) and k-Nearest Neighbors (kNN). We compare the performance of our model with that of the...
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.