Sharif Digital Repository

Twitter sentiment analysis, which often focuses on predicting the polarity of tweets, has attracted increasing attention over the last years, in particular with the rise of deep learning (DL). In this paper, we propose a new task: predicting the predominant sentiment among (first-order) replies to a given tweet. Therefore, we created RETwEET, a large dataset of tweets and replies manually annotated with sentiment labels. As a strong baseline, we propose a two-stage DL-based method: first, we create automatically labeled training data by applying a standard sentiment classifier to tweet replies and aggregating its predictions for each original tweet; our rationale is that individual errors... 

The orientation of a conducting target was estimated with a generalized regression neural network (GRNN) network using back-scattered data. Steps made to provide the training data sets by a computer code using method of moments were investigated. Noisy data sets were provided to improve the system generalization. The GRNN target orientation estimation and its performance was evaluated and the robustness was verified against target scaling, target deformation, sensor misplacements and introduction of noise with different S/N into sensor measurements  

The voluminous of the e-mails are spam. Several algorithms are represented for spam detection based on batch learning. In this paper, a new algorithm based on incremental learning is introduced. The algorithm composes new knowledge from new training data with previous knowledge by combining classifiers based on weighted majority voting. The experiment results show that the proposed algorithm outperforms other related incremental algorithms and non-incremental algorithms  

In this paper, we propose an ensemble-based approach to boost performance of Tied Factor Analysis(TFA) to overcome some of the challenges in face recognition across large pose variations. We use Adaboost.m1 to boost TFA which has shown to possess state-of-the-art face recognition performance under large pose variations. To this end, we have employed boosting as a discriminative training in the TFA as a generative model. In this model, TFA is used as a base classiœr for the boosting algorithm and a weighted likelihood model for TFA is proposed to adjust the importance of each training data. Moreover, a modiÔd weighting and a diversity criterion are used to generate more diverse classiœrs in... 

We propose an ℓ1 criterion for dictionary learning for sparse signal representation. Instead of directly searching for the dictionary vectors, our dictionary learning approach identifies vectors that are orthogonal to the subspaces in which the training data concentrate. We study conditions on the coefficients of training data that guarantee that ideal normal vectors deduced from the dictionary are local optima of the criterion. We illustrate the behavior of the criterion on a 2D example, showing that the local minima correspond to ideal normal vectors when the number of training data is sufficient. We conclude by describing an algorithm that can be used to optimize the criterion in higher... 

The purpose of dictionary learning problem is to learn a dictionary D from a training data matrix Y such that Y ≈ DX and the coefficient matrix X is sparse. Many algorithms have been introduced to this aim, which minimize the representation error subject to a sparseness constraint on X. However, the dictionary learning problem is non-convex with respect to the pair (D,X). In a previous work [Sadeghi et al., 2013], a convex approximation to the non-convex term DX has been introduced which makes the whole DL problem convex. This approach can be almost applied to any existing DL algorithm and obtain better algorithms. In the current paper, it is shown that a simple modification on that approach... 

Human action recognition by using standard video files is a well-studied problem in the literature. In this study, we assume to have access to single modality standard data of some actions (training data). Based on this data, we aim at identifying the actions that are present in a target modality video data without any explicit source–target relationship information. In this case, the training and test phases of the recognition task are based on different imaging modalities. Our goal in this paper is to introduce a mapping (a nonlinear operator) on both modalities such that the outcome shares some common features. These common features were then used to recognize the actions in each domain.... 

Dictionary learning (DL) for sparse representation has been widely investigated during the last decade. A DL algorithm uses a training data set to learn a set of basis functions over which all training signals can be sparsely represented. In practice, training signals may contain a few outlier data, whose structures differ from those of the clean training set. The presence of these unpleasant data may heavily affect the learning performance of a DL algorithm. In this paper we propose a robust-to-outlier formulation of the DL problem. We then present an algorithm for solving the resulting problem. Experimental results on both synthetic data and image denoising demonstrate the promising... 

Discriminative approaches to human pose estimation have became popular in recent years. These approaches face a big challenge: Similar inputs might correspond to very dissimilar poses. This property misleads the mapping functions which rely on the Euclidean distances in the input space. In this paper, we use the distances between the labels of the training data to learn a metric and map the input data to a space where this problem is minimized. Our mapping is linear and hence preserves the manifold structure of the input data. We benefit from the unlabeled data to estimate this manifold in the new space as a nearest neighbor graph. We finally utilize Tikhonov regularization to find a smooth... 

The first spoken dialogue system is developed for the Persian language is introduced. This is a ticket reservation system with Persian ASR and NLU modules. The focus of the paper is on learning the dialogue management module. In this work, real on-line training data are used during the learning process. For on-line learning, the effect of the variations of discount factor (γ) on the learning speed is investigated as the second contribution of the research. The optimal values for γ were found and the variation pattern of the action-value function (Q) in the learning process was obtained. A probabilistic policy for selecting actions is used in this work for the first time instead of greedy... 

In this paper, the artificial neural network has been applied to estimate the reaction rate of methanol dehydration in dimethyl ether synthesis. The multilayer feed forward neural network with three inputs and one output has been trained with different algorithms and different numbers of neurons in the hidden layer. Two thirds of all training data are used for training of the network and the rest are used for testing of the generalization of the network. The accuracy of the proposed model was found to agree well with the experimental results over a wide range of experimental conditions. The results clearly depict that the neural network is a powerful tool to estimate the reaction rate and... 

In this paper, Bayesian Network (BN) is used for reliability assessment of composite power systems with emphasis on the importance of system components. A simple approach is presented to construct the BN associated with a given power system. The approach is based on the capability of the BN to learn from data which makes it possible to be applied to large power systems. The required training data is provided by state sampling using the Monte Carlo simulation. The constructed BN is then used to perform different probabilistic assessments such as ranking the criticality and importance of system components from reliability perspective. The BN is also used to compute the frequency and... 

In this study, neural network models have been used to predict the mechanical behaviors of mouse embryos. In addition, sensitivity analysis has been carried out to investigate the influence of the significance of input parameters on the mechanical behavior of mouse embryos. In order to reach these purposes two neural network models have been implemented. Experimental data earlier deduced-by [Flückiger, M. (2004). Cell Membrane Mechanical Modeling for Microrobotic Cell Manipulation. Diploma Thesis, ETHZ Swiss Federal Institute of Technology, Zurich, WS03/04]-were collected to obtain training and test data for the neural network. The results of these investigations show that the correlation... 

In this paper, we introduce a novel method for human activity recognition that benefits from the structure and sequential properties of the test data as well as the training data. In the training phase, we obtain a fraction of data labels at constant time intervals and use them in a semi-supervised graph-based method for recognizing the user's activities. We use label propagation on a k-nearest neighbor graph to calculate the probability of association of the unlabeled data to each class in this phase. Then we use these probabilities to train an HMM in a way that each of its hidden states corresponds to one class of activity. These probabilities are used to learn the transition probabilities... 

Increasing the number of car accidents and other cerebral disease cause to progress in using Brain-Compute Interface (BCI) as a common subject for research and treatment. The aim of Brain-Computer Interface system is to establish a new communication system that translates human intentions, reflected by brain signals, into a control signal for an output device such as a computer. To this end, different processes must be done on brain signals and these signals must be classified by suitable methods. There are various methods to classify ECoG signals which are different in features and classifiers. Used features depend on extracted features, feature reduction methods and measures of feature... 

Searching large image databases is a time consuming process when done manually. Current CBIR methods mostly rely on training data in specific domains. When source and domain of images are unknown, unsupervised methods provide better solutions. In this work, we use a hierarchical clustering scheme to group images in an unknown and large image database. In addition, the user should provide the current class assignment of a small number of images as a feedback to the system. The proposed method uses this feedback to guess the number of required clusters, and optimizes the weight vector in an iterative manner. In each step, after modification of the weight vector, the images are reclustered. We... 

With recent deregulation in electricity industry, price forecasting has become the basis for this competitive market. The precision of this forecasting is essential in bidding strategies. So far, the artificial neural networks which can find an accurate relation between the historical data and the price have been used for this purpose. One major problem is that, they usually need a large number of training data and neurons either for complex function approximation and data fitting or classification and pattern recognition. As a result, the network topology has a significant impact on the network computational time and ability to learn and also to generate unseen data from training data. To... 

We present and evaluate Deep Private-Feature Extractor (DPFE), a deep model which is trained and evaluated based on information theoretic constraints. Using the selective exchange of information between a user's device and a service provider, DPFE enables the user to prevent certain sensitive information from being shared with a service provider, while allowing them to extract approved information using their model. We introduce and utilize the log-rank privacy, a novel measure to assess the effectiveness of DPFE in removing sensitive information and compare different models based on their accuracy-privacy trade-off. We then implement and evaluate the performance of DPFE on smartphones to... 

Nowadays, geo-distributed Data Centers (DCs) are very common, because of providing more energy efficiency, higher system availability as well as flexibility. In a geo-distributed cloud, each local DC responds to the specific portion of the incoming load which distributed based on different Geographically Load Balancing (GLB) policies. As a large yet flexible power consumer, the local DC has a great impact on the local power grid. From this point of view, a local DC is a good candidate to participate in the emerging power market such as Regulation Service (RS) opportunity, that brings monetary benefits both for the DC as well as the grid. However, a fruitful collaboration requires the DC to... 

Nowadays, geo-distributed Data Centers (DCs) are very common, because of providing more energy efficiency, higher system availability as well as flexibility. In a geo-distributed cloud, each local DC responds to the specific portion of the incoming load which distributed based on different Geographically Load Balancing (GLB) policies. As a large yet flexible power consumer, the local DC has a great impact on the local power grid. From this point of view, a local DC is a good candidate to participate in the emerging power market such as Regulation Service (RS) opportunity, that brings monetary benefits both for the DC as well as the grid. However, a fruitful collaboration requires the DC to...