Showing papers in "Journal of AI and Data Mining in 2020"
TL;DR: An improved community detection method called SD-GCN is proposed which uses a hybrid node scoring and synchronous label updating of boundary nodes, along with disabling random label updating in initial updates, and a new method for merging communities in second phase which is faster than modularity-based methods.
Abstract: Community structure is vital to discover the important structures and potential property of complex networks. In recent years, the increasing quality of local community detection approaches has become a hot spot in the study of complex network due to the advantages of linear time complexity and applicable for large-scale networks. However, there are many shortcomings in these methods such as instability, low accuracy, randomness, etc. The G-CN algorithm is one of local methods that uses the same label propagation as the LPA method, but unlike the LPA, only the labels of boundary nodes are updated at each iteration that reduces its execution time. However, it has resolution limit and low accuracy problem. To overcome these problems, this paper proposes an improved community detection method called SD-GCN which uses a hybrid node scoring and synchronous label updating of boundary nodes, along with disabling random label updating in initial updates. In the first phase, it updates the label of boundary nodes in a synchronous manner using the obtained score based on degree centrality and common neighbor measures. In addition, we defined a new method for merging communities in second phase which is faster than modularity-based methods. Extensive set of experiments are conducted to evaluate performance of the SD-GCN on small and large-scale real-world networks and artificial networks. These experiments verify significant improvement in the accuracy and stability of community detection approaches in parallel with shorter execution time in a linear time complexity.
8 citations
TL;DR: Evaluation results demonstrate that the proposed deep model, entitled Deep Block Super Resolution (DBSR), outperforms state-of-the-art alternatives in both areas of medical and general super-resolution enhancement from a single input image.
Abstract: This study presents a method to reconstruct a high-resolution image using a deep convolution neural network. We propose a deep model, entitled Deep Block Super Resolution (DBSR), by fusing the output features of a deep convolutional network and a shallow convolutional network. In this way, our model benefits from high frequency and low frequency features extracted from deep and shallow networks simultaneously. We use the residual layers in our model to make repetitive layers, increase the depth of the model, and make an end-to-end model. Furthermore, we employed a deep network in up-sampling step instead of bicubic interpolation method used in most of the previous works. Since the image resolution plays an important role to obtain rich information from the medical images and helps for accurate and faster diagnosis of the ailment, we use the medical images for resolution enhancement. Our model is capable of reconstructing a high-resolution image from low-resolution one in both medical and general images. Evaluation results on TSA and TZDE datasets, including MRI images, and Set5, Set14, B100, and Urban100 datasets, including general images, demonstrate that our model outperforms state-of-the-art alternatives in both areas of medical and general super-resolution enhancement from a single input image.
8 citations
TL;DR: Simulations on highway environment show that the proposed FLOR protocol has a better QoS efficiency compared to the above published methods in the literature.
Abstract: Vehicular ad hoc networks are an emerging technology with an extensive capability in various applications including vehicles safety, traffic management and intelligent transportation systems. Considering the high mobility of vehicles and their inhomogeneous distributions, designing an efficient routing protocol seems necessary. Given the fact that a road is crowded at some sections and is not crowded at the others, the routing protocol should be able to dynamically make decisions. On the other hand, VANET networks environment is vulnerable at the time of data transmission. Broadcast routing, similar to opportunistic routing, could offer better efficiency compared to other protocols. In this paper, a fuzzy logic opportunistic routing (FLOR) protocol is presented in which the packet rebroadcasting decision-making process is carried out through the fuzzy logic system along with three input parameters of packet advancement, local density, and the number of duplicated delivered packets. The rebroadcasting procedures use the value of these parameters as inputs to the fuzzy logic system to resolve the issue of multicasting, considering the crowded and sparse zones. NS-2 simulator is used for evaluating the performance of the proposed FLOR protocol in terms of packet delivery ratio, the end-to-end delay, and the network throughput compared with the existing protocols such as: FLOODING, P-PERSISTENCE and FUZZBR. The performance comparison also emphasizes on effective utilization of the resources. Simulations on highway environment show that the proposed protocol has a better QoS efficiency compared to the above published methods in the literature
7 citations
TL;DR: A sampling algorithm that equipped with an evaluator unit for analyzing the edges and a set of simple fixed structure learning automata and results show the superiority of the proposed algorithm are compared with the best current sampling algorithm.
Abstract: Social networks are streaming, diverse and include a wide range of edges so that continuously evolves over time and formed by the activities among users (such as tweets, emails, etc.), where each activity among its users, adds an edge to the network graph. Despite their popularities, the dynamicity and large size of most social networks make it difficult or impossible to study the entire network. This paper proposes a sampling algorithm that equipped with an evaluator unit for analyzing the edges and a set of simple fixed structure learning automata. Evaluator unit evaluates each edge and then decides whether edge and corresponding node should be added to the sample set. In The proposed algorithm, each main activity graph node is equipped with a simple learning automaton. The proposed algorithm is compared with the best current sampling algorithm that was reported in the Kolmogorov-Smirnov test (KS) and normalized L1 and L2 distances in real networks and synthetic networks presented as a sequence of edges. Experimental results show the superiority of the proposed algorithm.
7 citations
TL;DR: In the proposed method real-world video frames are used to determine the exact type of vehicle, and the accuracy of 89.5% is achieved, which represents a good performance.
Abstract: Fine-grained vehicle type recognition is one of the main challenges in machine vision. Almost all of the ways presented so far have identified the type of vehicle with the help of feature extraction and classifiers. Because of the apparent similarity between car classes, these methods may produce erroneous results. This paper presents a methodology that uses two criteria to identify common vehicle types. The first criterion is feature extraction and classification and the second criterion is to use the dimensions of car for classification. This method consists of three phases. In the first phase, the coordinates of the vanishing points are obtained. In the second phase, the bounding box and dimensions are calculated for each passing vehicle. Finally, in the third phase, the exact vehicle type is determined by combining the results of the first and second criteria. To evaluate the proposed method, a dataset of images and videos, prepared by the authors, has been used. This dataset is recorded from places similar to those of a roadside camera. Most existing methods use high-quality images for evaluation and are not applicable in the real world, but in the proposed method real-world video frames are used to determine the exact type of vehicle, and the accuracy of 89.5% is achieved, which represents a good performance.
7 citations
TL;DR: A Reliable Controller Placement Problem Model (RCPPM) is proposed to solve the NP-hard RCPPM in a heuristic manner, and performance evaluations show the efficiency of the proposed framework.
Abstract: Software-Defined Network (SDNs) is a decoupled architecture that enables administrators to build a customizable and manageable network. Although the decoupled control plane provides flexible management and facilitates the task of operating the network, it is the vulnerable point of failure in SDN. To achieve a reliable control plane, multiple controller are often needed so that each switch must be assigned to more than one controller. In this paper, a Reliable Controller Placement Problem Model (RCPPM) is proposed to solve such a problem, so as to maximize the reliability of software defined networks. Unlike previous works that only consider latencies parameters, the new model takes into account the load of control traffic and reliability metrics as well. Furthermore, a near-optimal algorithm is proposed to solve the NP-hard RCPPM in a heuristic manner. Finally, through extensive simulation, a comprehensive analysis of the RCPPM is presented for various topologies extracted from Internet Topology Zoo. Our performance evaluations show the efficiency of the proposed framework.
6 citations
TL;DR: This paper suggests an effective solution method based on the so-called Iterated Local Search (ILS) strategy that is computationally much more effective and efficient over middle to large instances of the controller placement problem.
Abstract: Software defined network is a new computer network architecture who separates controller and data layer in network devices such as switches and routers. By the emerge of software defined networks, a class of location problems, called controller placement problem, has attracted much more research attention. The task in the problem is to simultaneously find optimal number and location of controllers satisfying a set of routing and capacity constraints. In this paper, we suggest an effective solution method based on the so-called Iterated Local Search (ILS) strategy. We then, compare our method to an existing standard mathematical programming solver on an extensive set of problem instances. It turns out that our suggested method is computationally much more effective and efficient over middle to large instances of the problem.
5 citations
TL;DR: A comparative study of experimental results embosses various advantages of the proposed technique such as accurate representation, low approximation errors and efficient computational complexity.
Abstract: In this paper, a new technique has been designed to capture the outline of 2D shapes using cubic B´ezier curves. The proposed technique avoids the traditional method of optimizing the global squared fitting error and emphasizes the local control of data points. A maximum error has been determined to preserve the absolute fitting error less than a criterion and it administers the process of curve subdivision. Depending on the specified maximum error, the proposed technique itself subdivides complex segments, and curve fitting is done simultaneously. A comparative study of experimental results embosses various advantages of the proposed technique such as accurate representation, low approximation errors and efficient computational complexity.
5 citations
TL;DR: This approach replaces a full state space generation, only by producing part of it checking the safety, and finding errors (e.g., deadlock), and shows that the proposed approach is more efficient and accurate compared to other approaches.
Abstract: Model checking is an automatic technique for software verification through which all reachable states are generated from an initial state to finding errors and desirable patterns. In the model checking approach, the behavior and structure of system should be modeled. Graph transformation system is a graphical formal modeling language to specify and model the system. However, modeling of large systems with the graph transformation system suffers from the state space explosion problem which usually requires huge amounts of computational resources. In this paper, we propose a hybrid meta-heuristic approach to deal with this searching problem in the graph transformation system because meta-heuristic algorithms are efficient solutions to traverse the graph of large systems. Our approach, using Artificial Bee Colony and Simulated Annealing, replaces a full state space generation, only by producing part of it checking the safety, and finding errors (e.g., deadlock). The experimental results show that our proposed approach is more efficient and accurate compared to other approaches.
5 citations
TL;DR: “starting from Accelerators,” “creativity and problem solving ability of founders”, “fist mover advantage” and “amount of seed investment” are the four most important variables which affects the start-ups success and the other 15 variables are less important.
Abstract: The purpose of this study is to reduce the uncertainty of early stage startups success prediction and filling the gap of previous studies in the field, by identifying and evaluating the success variables and developing a novel business success failure (S/F) data mining classification prediction model for Iranian start-ups. For this purpose, the paper is seeking to extend Bill Gross and Robert Lussier S/F prediction model variables and algorithms in a new context of Iranian start-ups which starts from accelerators in order to build a new S/F prediction model. A sample of 161 Iranian start-ups which are based in accelerators from 2013 to 2018 is applied and 39 variables are extracted from the literature and organized in five groups. Then the sample is fed into six well-known classification algorithms. Two staged stacking as a classification model is the best performer among all other six classification based S/F prediction models and it can predict binary dependent variable of success or failure with accuracy of 89% on average. Also finding shows that “starting from Accelerators”, “creativity and problem solving ability of founders”, “fist mover advantage” and “amount of seed investment” are the four most important variables which affects the start-ups success and the other 15 variables are less important.
4 citations
TL;DR: An ensemble multi-stage machine for survivability prediction which obtains considerable accuracy while it ignores difficult features for most of the input samples and is evaluated using the Surveillance, Epidemiology, and End Results database.
Abstract: Prediction of cancer survivability using machine learning techniques has become a popular approach in recent years. In this regard, an important issue is that preparation of some features may need conducting difficult and costly experiments while these features have less significant impacts on the final decision and can be ignored from the feature set. Therefore, developing a machine for prediction of survivability, which ignores these features for simple cases and yields an acceptable prediction accuracy, has turned into a challenge for researchers. In this paper, we have developed an ensemble multi-stage machine for survivability prediction which ignores difficult features for simple cases. The machine employs three basic learners, namely multilayer perceptron (MLP), support vector machine (SVM), and decision tree (DT), in the first stage to predict survivability using simple features. If the learners agree on the output, the machine makes the final decision in the first stage. Otherwise, for difficult cases where the output of learners is different, the machine makes decision in the second stage using SVM over all features. The developed model was evaluated using the Surveillance, Epidemiology, and End Results (SEER) database. The experimental results revealed that the developed machine obtains considerable accuracy while it ignores difficult features for most of the input samples.
TL;DR: Shuffled Frog Leaping Programming (SFLP) inspired by SFLA is proposed as a novel type of automatic programming model to solve symbolic regression problems based on tree representation and a new mechanism for improving constant numbers in the tree structure is proposed.
Abstract: There are various automatic programming models inspired by evolutionary computation techniques. Due to the importance of devising an automatic mechanism to explore the complicated search space of mathematical problems where numerical methods fails, evolutionary computations are widely studied and applied to solve real world problems. One of the famous algorithm in optimization problem is shuffled frog leaping algorithm (SFLA) which is inspired by behaviour of frogs to find the highest quantity of available food by searching their environment both locally and globally. The results of SFLA prove that it is competitively effective to solve problems. In this paper, Shuffled Frog Leaping Programming (SFLP) inspired by SFLA is proposed as a novel type of automatic programming model to solve symbolic regression problems based on tree representation. Also, in SFLP, a new mechanism for improving constant numbers in the tree structure is proposed. In this way, different domains of mathematical problems can be addressed with the use of proposed method. To find out about the performance of generated solutions by SFLP, various experiments were conducted using a number of benchmark functions. The results were also compared with other evolutionary programming algorithms like BBP, GSP, GP and many variants of GP.
TL;DR: A semantic labeling method is proposed by fusion of optical and normalized DSM data to train the classifier and shows significant improvement in tree, building, and car accuracy.
Abstract: Semantic labeling is an active field in remote sensing applications. Although handling high detailed objects in Very High Resolution (VHR) optical image and VHR Digital Surface Model (DSM) is a challenging task, it can improve the accuracy of semantic labeling methods. In this paper, a semantic labeling method is proposed by fusion of optical and normalized DSM data. Spectral and spatial features are fused into a Heterogeneous Feature Map to train the classifier. Evaluation database classes are impervious surface, building, low vegetation, tree, car, and background. The proposed method is implemented on Google Earth Engine. The method consists of several levels. First, Principal Component Analysis is applied to vegetation indexes to find maximum separable color space between vegetation and non-vegetation area. Gray Level Co-occurrence Matrix is computed to provide texture information as spatial features. Several Random Forests are trained with automatically selected train dataset. Several spatial operators follow the classification to refine the result. Leaf-Less-Tree feature is used to solve the underestimation problem in tree detection. Area, major and, minor axis of connected components are used to refine building and car detection. Evaluation shows significant improvement in tree, building, and car accuracy. Overall accuracy and Kappa coefficient are appropriate.
TL;DR: Results show that applying the proposed method significantly reduces the misclassification cost by at least 14% compared with Decision tree, Naive bayes, Bayesian Network, Neural network and Artificial immune system.
Abstract: Due to today’s advancement in technology and businesses, fraud detection has become a critical component of financial transactions. Considering vast amounts of data in large datasets, it becomes more difficult to detect fraud transactions manually. In this research, we propose a combined method using both data mining and statistical tasks, utilizing feature selection, resampling and cost-sensitive learning for credit card fraud detection. In the first step, useful features are identified using genetic algorithm. Next, the optimal resampling strategy is determined based on the design of experiments (DOE) and response surface methodologies. Finally, the cost sensitive C4.5 algorithm is used as the base learner in the Adaboost algorithm. Using a real-time data set, results show that applying the proposed method significantly reduces the misclassification cost by at least 14% compared with Decision tree, Naive bayes, Bayesian Network, Neural network and Artificial immune system.
TL;DR: A dataset including bibliographic information of user’s publication, the publication venues, and other published papers provided as a way to find an expert in a particular context where experts are recommended to a user according to his records and preferences is provided.
Abstract: Online scientific communities are bases that publish books, journals, and scientific papers, and help promote knowledge. The researchers use search engines to find the given information including scientific papers, an expert to collaborate with, and the publication venue, but in many cases due to search by keywords and lack of attention to the content, they do not achieve the desired results at the early stages. Online scientific communities can increase the system efficiency to respond to their users utilizing a customized search. In this paper, using a dataset including bibliographic information of user’s publication, the publication venues, and other published papers provided as a way to find an expert in a particular context where experts are recommended to a user according to his records and preferences. In this way, a user request to find an expert is presented with keywords that represent a certain expertise and the system output will be a certain number of ranked suggestions for a specific user. Each suggestion is the name of an expert who has been identified appropriate to collaborate with the user. In evaluation using IEEE database, the proposed method reached an accuracy of 71.50 percent that seems to be an acceptable result.
TL;DR: The results confirm that appropriate usage of 2D CNNs outperforms a 3D CNN implementation in this task, and efficiently combined color and depth modalities and 2D and 3DCNN predictions.
Abstract: Despite considerable enhances in recognizing hand gestures from still images, there are still many challenges in the classification of hand gestures in videos. The latter comes with more challenges, including higher computational complexity and arduous task of representing temporal features. Hand movement dynamics, represented by temporal features, have to be extracted by analyzing the total frames of a video. So far, both 2D and 3D convolutional neural networks have been used to manipulate the temporal dynamics of the video frames. 3D CNNs can extract the changes in the consecutive frames and tend to be more suitable for the video classification task, however, they usually need more time. On the other hand, by using techniques like tiling it is possible to aggregate all the frames in a single matrix and preserve the temporal and spatial features. This way, using 2D CNNs, which are inherently simpler than 3D CNNs can be used to classify the video instances. In this paper, we compared the application of 2D and 3D CNNs for representing temporal features and classifying hand gesture sequences. Additionally, providing a two-stage two-stream architecture, we efficiently combined color and depth modalities and 2D and 3D CNN predictions. The effect of different types of augmentation techniques is also investigated. Our results confirm that appropriate usage of 2D CNNs outperforms a 3D CNN implementation in this task.
TL;DR: The results of investigating 7 Dimensions and 21 Criteria show that the quality of teaching this course increases, if the updated teaching methods and contents to be used, the evaluation policy to be tailored to the course, the course professor and his/her assistants be available to correct students' mistakes and there is also an interactive system based on student comments.
Abstract: Identification of the factors affecting teaching quality of engineering drawing and interaction between them is necessary until it is determined which manipulation will improve the quality of teaching this course. Since the above issue is a Multi-Criteria Decision Making (MCDM) problem and on the other hand, we are faced with human factors, the Fuzzy DEMATEL method is suggested for solving it. Also, because DEMATEL analysis does not lead to a weighting of the criteria, it is combined with the ANP and a hybrid fuzzy DEMATEL-ANP (FDANP) methodology is used. The results of investigating 7 Dimensions and 21 Criteria show that the quality of teaching this course increases, if the updated teaching methods and contents to be used, the evaluation policy to be tailored to the course, the course professor and his/her assistants be available to correct students' mistakes and there is also an interactive system based on student comments.
TL;DR: An enhancement of the fuzzy c-means clustering algorithm is proposed which is suitable for clustering trapezoidal fuzzy data and substitutes missing attribute by a trapezoid fuzzy number to be determined by using the corresponding attribute of q nearest-neighbor.
Abstract: The fuzzy c-means clustering algorithm is a useful tool for clustering; but it is convenient only for crisp complete data. In this article, an enhancement of the algorithm is proposed which is suitable for clustering trapezoidal fuzzy data. A linear ranking function is used to define a distance for trapezoidal fuzzy data. Then, as an application, a method based on the proposed algorithm is presented to cluster incomplete fuzzy data. The method substitutes missing attribute by a trapezoidal fuzzy number to be determined by using the corresponding attribute of q nearest-neighbor. Comparisons and analysis of the experimental results demonstrate the capability of the proposed method.
TL;DR: Simulation results show that the proposed algorithm based on the learned dictionaries derived from the combinational features can detect the type of rice grain and determine its quality precisely.
Abstract: In scientific and commercial fields associated with modern agriculture, the categorization of different rice types and determination of its quality is very important. Various image processing algorithms are applied in recent years to detect different agricultural products. The problem of rice classification and quality detection in this paper is presented based on model learning concepts including sparse representation and dictionary learning techniques to yield over-complete models in this processing field. There are color-based, statistical-based and texture-based features to represent the structural content of rice varieties. To achieve the desired results, different features from recorded images are extracted and used to learn the representative models of rice samples. Also, sparse principal component analysis and sparse structured principal component analysis is employed to reduce the dimension of classification problem and lead to an accurate detector with less computational time. The results of the proposed classifier based on the learned models are compared with the results obtained from neural network and support vector machine. Simulation results, along with a meaningful statistical test, show that the proposed algorithm based on the learned dictionaries derived from the combinational features can detect the type of rice grain and determine its quality precisely.
TL;DR: This paper presents an image denoising method based on a maximum a posteriori (MAP) density function estimator, which is implemented in the wavelet domain because of its energy compaction property.
Abstract: Removing noise from images is a challenging problem in digital image processing. This paper presents an image denoising method based on a maximum a posteriori (MAP) density function estimator, which is implemented in the wavelet domain because of its energy compaction property. The performance of the MAP estimator depends on the proposed model for noise-free wavelet coefficients. Thus in the wavelet based image denoising, selecting a proper model for wavelet coefficients is very important. In this paper, we model wavelet coefficients in each sub-band by heavy-tail distributions that are from scale mixture of normal distribution family. The parameters of distributions are estimated adaptively to model the correlation between the coefficient amplitudes, so the intra-scale dependency of wavelet coefficients is also considered. The denoising results confirm the effectiveness of the proposed method.
TL;DR: The proposed algorithm uses three objective functions to achieve high-quality discretization and introduces a new criterion called the normalized cut, which uses the relationships between their features’ values to maintain the nature of the data.
Abstract: Learning models and related results depend on the quality of the input data. If raw data is not properly cleaned and structured, the results are tending to be incorrect. Therefore, discretization as one of the preprocessing techniques plays an important role in learning processes. The most important challenge in the discretization process is to reduce the number of features’ values. This operation should be applied in a way that relationships between the features are maintained and accuracy of the classification algorithms would increase. In this paper, a new evolutionary multi-objective algorithm is presented. The proposed algorithm uses three objective functions to achieve high-quality discretization. The first and second objectives minimize the number of selected cut points and classification error, respectively. The third objective introduces a new criterion called the normalized cut, which uses the relationships between their features’ values to maintain the nature of the data. The performance of the proposed algorithm was tested using 20 benchmark datasets. According to the comparisons and the results of nonparametric statistical tests, the proposed algorithm has a better performance than other existing major methods.
TL;DR: The empirical results demonstrate the superiority of the proposed chaotic AIW-PSO to the counterparts over 21 functions, which confirms the promising role of inserting the randomness into the AIW -PSO.
Abstract: Among variety of meta-heuristic population-based search algorithms, particle swarm optimization (PSO) with adaptive inertia weight (AIW) has been considered as a versatile optimization tool, which incorporates the experience of the whole swarm into the movement of particles. Although the exploitation ability of this algorithm is great, it cannot comprehensively explore the search space and may be trapped in a local minimum through a limited number of iterations. To increase its diversity as well as enhancing its exploration ability, this paper inserts a chaotic factor, generated by three chaotic systems, along with a perturbation stage into AIW-PSO to avoid premature convergence, especially in complex nonlinear problems. To assess the proposed method, a known optimization benchmark containing nonlinear complex functions was selected and its results were compared to that of standard PSO, AIW-PSO and genetic algorithm (GA). The empirical results demonstrate the superiority of the proposed chaotic AIW-PSO to the counterparts over 21 functions, which confirms the promising role of inserting the randomness into the AIW-PSO. The behavior of error through the epochs show that the proposed manner can smoothly find proper minimums in a timely manner without encountering with premature convergence.
TL;DR: A several syntactic features based on dependency grammar along with some morphological and language-independent features have been designed in order to extract suitable features for the learning phase in Conditional Random Field-based Persian Named Entity Recognition.
Abstract: Named Entity Recognition is an information extraction technique that identifies name entities in a text. Three popular methods have been conventionally used namely: rule-based, machine-learning-based and hybrid of them to extract named entities from a text. Machine-learning-based methods have good performance in the Persian language if they are trained with good features. To get good performance in Conditional Random Field-based Persian Named Entity Recognition, a several syntactic features based on dependency grammar along with some morphological and language-independent features have been designed in order to extract suitable features for the learning phase. In this implementation, designed features have been applied to Conditional Random Field to build our model. To evaluate our system, the Persian syntactic dependency Treebank with about 30,000 sentences, prepared in NOOR Islamic science computer research center, has been implemented. This Treebank has Named-Entity tags, such as Person, Organization and location. The result of this study showed that our approach achieved 86.86% precision, 80.29% recall and 83.44% F-measure which are relatively higher than those values reported for other Persian NER methods.
TL;DR: A two-phase method for removing salt and pepper noise while preserving edges and fine details is proposed and outperforms to several existing methods.
Abstract: Removing salt and pepper noise is an active research area in image processing. In this paper, a two-phase method is proposed for removing salt and pepper noise while preserving edges and fine details. In the first phase, noise candidate pixels are detected which are likely to be contaminated by noise. In the second phase, only noise candidate pixels are restored using adaptive median filter. In terms of noise detection, a two-stage method is utilized. At first, a thresholding is applied on the image to initial estimation of the noise candidate pixels. Since some pixels in the image may be similar to the salt and pepper noise, these pixels are mistakenly identified as noise. Hence, in the second step of the noise detection, the pixon-based segmentation is used to identify the salt and pepper noise pixels more accurately. Pixon is the neighboring pixels with similar gray levels. The proposed method was evaluated on several noisy images, and the results show the accuracy of the proposed method in salt and pepper noise removal and outperforms to several existing methods.
TL;DR: A different reinforcement-based optimization approach called LA-OMF was proposed to find both the number and positions of TMFs for fuzzy association rules, and improved the efficiency of mining fuzzy associationrules by extracting optimized membership functions.
Abstract: The Transactions in web data often consist of quantitative data, suggesting that fuzzy set theory can be used to represent such data. The time spent by users on each web page is one type of web data, was regarded as a trapezoidal membership function (TMF) and can be used to evaluate user browsing behavior. The quality of mining fuzzy association rules depends on membership functions and since the membership functions of each web page are different from those of other web pages, so automatic finding the number and position of TMF is significant. In this paper, a different reinforcement-based optimization approach called LA-OMF was proposed to find both the number and positions of TMFs for fuzzy association rules. In the proposed algorithm, the centers and spreads of TMFs were considered as parameters of the search space, and a new representation using learning automata (LA) was proposed to optimize these parameters. The performance of the proposed approach was evaluated and the results were compared with the results of other algorithms on a real dataset. Experiments on datasets with different sizes confirmed that the proposed LA-OMF improved the efficiency of mining fuzzy association rules by extracting optimized membership functions.
TL;DR: A new method for decreasing the error of EEG- based key generation process based on window segmentation protocol, which obtains 0.76%, and 0.48% mean Half Total Error Rate (HTER) for 18-channel and single-channel cryptographic key generation systems respectively.
Abstract: Network security is very important when sending confidential data through the network. Cryptography is the science of hiding information, and a combination of cryptography solutions with cognitive science starts a new branch called cognitive cryptography that guarantee the confidentiality and integrity of the data. Brain signals as a biometric indicator can convert to a binary code which can be used as a cryptographic key. This paper proposes a new method for decreasing the error of EEG- based key generation process. Discrete Fourier Transform, Discrete Wavelet Transform, Autoregressive Modeling, Energy Entropy, and Sample Entropy were used to extract features. All features are used as the input of new method based on window segmentation protocol then are converted to the binary mode. We obtain 0.76%, and 0.48% mean Half Total Error Rate (HTER) for 18-channel and single-channel cryptographic key generation systems respectively.
TL;DR: A robust method for multi-view face detection in open environments, using a combination of Gabor features and neural networks, is presented and achieves great detection accuracy, by comparing it with several popular face-detection algorithms, such as OpenCV’s Viola-Jones detector.
Abstract: Multi-view face detection in open environments is a challenging task, due to the wide variations in illumination, face appearances and occlusion. In this paper, a robust method for multi-view face detection in open environments, using a combination of Gabor features and neural networks, is presented. Firstly, the effect of changing the Gabor filter parameters (orientation, frequency, standard deviation, aspect ratio and phase offset) for an image is analysed, secondly, the range of Gabor filter parameter values is determined and finally, the best values for these parameters are specified. A multilayer feedforward neural network with a back-propagation algorithm is used as a classifier. The input vector is obtained by convolving the input image and a Gabor filter, with both the angle and frequency values equal to π/2. The proposed algorithm is tested on 1,484 image samples with simple and complex backgrounds. The experimental results show that the proposed detector achieves great detection accuracy, by comparing it with several popular face-detection algorithms, such as OpenCV’s Viola-Jones detector.
TL;DR: A web service recommendation method called Popular-Dependent Collaborative Filtering (PDCF) is proposed which handles QoS differences experienced by the users as well as the dependency among users on a specific web service using the user/web service dependency factor.
Abstract: Since, most of the organizations present their services electronically, the number of functionally-equivalent web services is increasing as well as the number of users that employ those web services. Consequently, plenty of information is generated by the users and the web services that lead to the users be in trouble in finding their appropriate web services. Therefore, it is required to provide a recommendation method for predicting the quality of web services (QoS) and recommending web services. Most of the existing collaborative filtering approaches don’t operate efficiently in recommending web services due to ignoring some effective factors such as dependency among users/web services, the popularity of users/web services, and the location of web services/users. In this paper, a web service recommendation method called Popular-Dependent Collaborative Filtering (PDCF) is proposed. The proposed method handles QoS differences experienced by the users as well as the dependency among users on a specific web service using the user/web service dependency factor. Additionally, the user/web service popularity factor is considered in the PDCF that significantly enhances its effectiveness. We also proposed a location-aware method called LPDCF which considers the location of web services into the recommendation process of the PDCF. A set of experiments is conducted to evaluate the performance of the PDCF and investigating the impression of the matrix factorization model on the efficiency of the PDCF with two real-world datasets. The results indicate that the PDCF outperforms other competing methods in most cases.
TL;DR: The rough-neural network (R-NN) is utilized for the identification of CRK without the usage of MISO structures, a stochastic gradient descent learning algorithm is proposed for training the R-NNs, and simulation results show the effectiveness of proposed methodology.
Abstract: Because of the existing interactions among the variables of a multiple input-multiple output (MIMO) nonlinear system, its identification is a difficult task, particularly in the presence of uncertainties. Cement rotary kiln (CRK) is a MIMO nonlinear system in the cement factory with a complicated mechanism and uncertain disturbances. The identification of CRK is very important for different purposes such as prediction, fault detection, and control. In the previous works, CRK was identified after decomposing it into several multiple input-single output (MISO) systems. In this paper, for the first time, the rough-neural network (R-NN) is utilized for the identification of CRK without the usage of MISO structures. R-NN is a neural structure designed on the base of rough set theory for dealing with the uncertainty and vagueness. In addition, a stochastic gradient descent learning algorithm is proposed for training the R-NNs. The simulation results show the effectiveness of proposed methodology.
TL;DR: This paper presents analytical relations for fast estimation of the embedded uncertainty in depth acquisition and then these relations, along with the 3D sampling arrangement are employed to define a cost function and demonstrates significant improvement in depth uncertainty in the achieved depth maps.
Abstract: In this paper we address the problem of automatic arrangement of cameras in a 3D system to enhance the performance of depth acquisition procedure. Lacking ground truth or a priori information, a measure of uncertainty is required to assess the quality of reconstruction. The mathematical model of iso-disparity surfaces provides an efficient way to estimate the depth estimation uncertainty which is believed to be related to the baseline length, focal length, panning angle and the pixel resolution in a stereo vision system. Accordingly, we first present analytical relations for fast estimation of the embedded uncertainty in depth acquisition and then these relations, along with the 3D sampling arrangement are employed to define a cost function. The optimal camera arrangement will be determined by minimizing the cost function with respect to the system parameters and the required constraints. Finally, the proposed algorithm is implemented on some 3D models. The simulation results demonstrate significant improvement (up to 35%) in depth uncertainty in the achieved depth maps compared with the traditional rectified camera setup.