Seyedeh H. Erfani
Abstract
Facial expressions are part of human language and are often used to convey emotions. Since humans are very different in their emotional representation through various media, the recognition of facial expression becomes a challenging problem in machine learning methods. Emotion and sentiment analysis ...
Read More
Facial expressions are part of human language and are often used to convey emotions. Since humans are very different in their emotional representation through various media, the recognition of facial expression becomes a challenging problem in machine learning methods. Emotion and sentiment analysis also have become new trends in social media. Deep Convolutional Neural Network (DCNN) is one of the newest learning methods in recent years that model a human's brain. DCNN achieves better accuracy with big data such as images. In this paper an automatic facial expression recognition (FER) method using the deep convolutional neural network is proposed. In this work, a way is provided to overcome the overfitting problem in training the deep convolutional neural network for FER, and also an effective pre-processing phase is proposed that is improved the accuracy of facial expression recognition. Here the results for recognition of seven emotional states (neutral, happiness, sadness, surprise, anger, fear, disgust) have been presented by applying the proposed method on the two largely used public datasets (JAFFE and CK+). The results show that in the proposed method, the accuracy of the FER is better than traditional FER methods and is about 98.59% and 96.89% for JAFFE and CK+ datasets, respectively.
F. Amiri; S. Abbasi; M. Babaie mohamadeh
Abstract
During the COVID-19 crisis, we face a wide range of thoughts, feelings, and behaviors on social media that play a significant role in spreading information regarding COVID-19. Trustful information, together with hopeful messages, could be used to control people's emotions and reactions during pandemics. ...
Read More
During the COVID-19 crisis, we face a wide range of thoughts, feelings, and behaviors on social media that play a significant role in spreading information regarding COVID-19. Trustful information, together with hopeful messages, could be used to control people's emotions and reactions during pandemics. This study examines Iranian society's resilience in the face of the Corona crisis and provides a strategy to promote resilience in similar situations. It investigates posts and news related to the COVID-19 pandemic in Iran, to determine which messages and references have caused concern in the community, and how they could be modified? and also which references were the most trusted publishers? Social network analysis methods such as clustering have been used to analyze data. In the present work, we applied a two-stage clustering method constructed on the self-organizing map and K-means. Because of the importance of social trust in accepting messages, This work examines public trust in social posts. The results showed trust in the health-related posts was less than social-related and cultural-related posts. The trusted posts were shared on Instagram and news sites. Health and cultural posts with negative polarity affected people's trust and led to negative emotions such as fear, disgust, sadness, and anger. So, we suggest that non-political discourses be used to share topics in the field of health.
A. Alijamaat; A. Reza NikravanShalmani; P. Bayat
Abstract
Multiple Sclerosis (MS) is a disease that destructs the central nervous system cell protection, destroys sheaths of immune cells, and causes lesions. Examination and diagnosis of lesions by specialists is usually done manually on Magnetic Resonance Imaging (MRI) images of the brain. Factors such as small ...
Read More
Multiple Sclerosis (MS) is a disease that destructs the central nervous system cell protection, destroys sheaths of immune cells, and causes lesions. Examination and diagnosis of lesions by specialists is usually done manually on Magnetic Resonance Imaging (MRI) images of the brain. Factors such as small sizes of lesions, their dispersion in the brain, similarity of lesions to some other diseases, and their overlap can lead to the misdiagnosis. Automatic image detection methods as auxiliary tools can increase the diagnosis accuracy. To this end, traditional image processing methods and deep learning approaches have been used. Deep Convolutional Neural Network is a common method of deep learning to detect lesions in images. In this network, the convolution layer extracts the specificities; and the pooling layer decreases the specificity map size. The present research uses the wavelet-transform-based pooling. In addition to decomposing the input image and reducing its size, the wavelet transform highlights sharp changes in the image and better describes local specificities. Therefore, using this transform can improve the diagnosis. The proposed method is based on six convolutional layers, two layers of wavelet pooling, and a completely connected layer that had a better amount of accuracy than the studied methods. The accuracy of 98.92%, precision of 99.20%, and specificity of 98.33% are obtained by testing the image data of 38 patients and 20 healthy individuals.
Mohammad Reza Keyvanpour; Zahra Karimi Zandian; Nasrin Mottaghi
Abstract
Regression testing reduction is an essential phase in software testing. In this step, the redundant and unnecessary cases are eliminated, whereas software accuracy and performance are not degraded. So far, various researches have been proposed in regression testing reduction field. The main challenge ...
Read More
Regression testing reduction is an essential phase in software testing. In this step, the redundant and unnecessary cases are eliminated, whereas software accuracy and performance are not degraded. So far, various researches have been proposed in regression testing reduction field. The main challenge in this area is to provide a method that maintain fault-detection capability while reducing test suites. In this paper, a new test suite reduction technique is proposed based on data mining. In this method, in addition to test suite reduction, its fault-detection capability is preserved using both clustering and classification. In this approach, regression test cases are reduced using a bi-criteria data mining-based method in two levels. In each level, the different and useful coverage criteria and clustering algorithms are used to establish a better compromise between test suite size and the ability of reduced test suite fault detection. The results of the proposed method have been compared to the effects of five other methods based on PSTR and PFDL. The experiments show the efficiency of the proposed method in the test suite reduction in maintaining its capability in fault detection.
Seyedeh S. Sadeghi; H. Khotanlou; M. Rasekh Mahand
Abstract
In the modern age, written sources are rapidly increasing. A growing number of these data are related to the texts containing the feelings and opinions of the users. Thus, reviewing and analyzing of emotional texts have received a particular attention in recent years. A System which is based on combination ...
Read More
In the modern age, written sources are rapidly increasing. A growing number of these data are related to the texts containing the feelings and opinions of the users. Thus, reviewing and analyzing of emotional texts have received a particular attention in recent years. A System which is based on combination of cognitive features and deep neural network, Gated Recurrent Unit has been proposed in this paper. Five basic emotions used in this approach are: anger, happiness, sadness, surprise and fear. A total of 23,000 Persian documents by the average length of 24 have been labeled for this research. Emotional constructions, emotional keywords, and emotional POS are the basic cognitive features used in this approach. On the other hand, after preprocessing the texts, words of normalized text have been embedded by Word2Vec technique. Then, a deep learning approach has been done based on this embedded data. Finally, classification algorithms such as Naïve Bayes, decision tree, and support vector machines were used to classify emotions based on concatenation of defined cognitive features, and deep learning features. 10-fold cross validation has been used to evaluate the performance of the proposed system. Experimental results show the proposed system achieved the accuracy of 97%. Result of proposed system shows the improvement of several percent’s in comparison by other results achieved GRU and cognitive features in isolation. At the end, studying other statistical features and improving these cognitive features in more details can affect the results.
V. Fazel Asl; B. Karasfi; B. Masoumi
Abstract
In this article, we consider the problems of abnormal behavior detection in a high-crowded environment. One of the main issues in abnormal behavior detection is the complexity of the structure patterns between the frames. In this paper, social force and optical flow patterns are used to prepare the system ...
Read More
In this article, we consider the problems of abnormal behavior detection in a high-crowded environment. One of the main issues in abnormal behavior detection is the complexity of the structure patterns between the frames. In this paper, social force and optical flow patterns are used to prepare the system for training the complexity of the structural patterns. The cycle GAN system has been used to train behavioral patterns. Two models of normal and abnormal behavioral patterns are used to evaluate the accuracy of the system detection. In the case of abnormal patterns used for training, due to the lack of this type of behavioral pattern, which is another challenge in detecting the abnormal behaviors, the geometric techniques are used to augment the patterns. If the normal behavioral patterns are used for training, there is no need to augment the patterns because the normal patterns are sufficient. Then, by using the cycle generative adversarial nets (cycle GAN), the normal and abnormal behaviors training will be considered separately. This system produces the social force and optical flow pattern for normal and abnormal behaviors on the first and second sides. We use the cycle GAN system both to train behavioral patterns and to assess the accuracy of abnormal behaviors detection. In the testing phase, if normal behavioral patterns are used for training, the cycle GAN system should not be able to reconstruct the abnormal behavioral patterns with high accuracy.
M. Asgari-Bidhendi; B. Janfada; O. R. Roshani Talab; B. Minaei-Bidgoli
Abstract
Named Entity Recognition (NER) is one of the essential prerequisites for many natural language processing tasks. All public corpora for Persian named entity recognition, such as ParsNERCorp and ArmanPersoNERCorpus, are based on the Bijankhan corpus, which is originated from the Hamshahri newspaper in ...
Read More
Named Entity Recognition (NER) is one of the essential prerequisites for many natural language processing tasks. All public corpora for Persian named entity recognition, such as ParsNERCorp and ArmanPersoNERCorpus, are based on the Bijankhan corpus, which is originated from the Hamshahri newspaper in 2004. Correspondingly, most of the published named entity recognition models in Persian are specially tuned for the news data and are not flexible enough to be applied in different text categories, such as social media texts. This study introduces ParsNER-Social, a corpus for training named entity recognition models in the Persian language built from social media sources. This corpus consists of 205,373 tokens and their NER tags, crawled from social media contents, including 10 Telegram channels in 10 different categories. Furthermore, three supervised methods are introduced and trained based on the ParsNER-Social corpus: Two conditional random field models as baseline models and one state-of-the-art deep learning model with six different configurations are evaluated on the proposed dataset. The experiments show that the Mono-Lingual Persian models based on Bidirectional Encoder Representations from Transformers (MLBERT) outperform the other approaches on the ParsNER-Social corpus. Among different Configurations of MLBERT models, the ParsBERT+BERT-TokenClass model obtained an F1-score of 89.65%.
Z. Nazari; H.R. Koohi; J. Mousavi
Abstract
Nowadays, with the expansion of the internet and its associated technologies, recommender systems have become increasingly common. In this work, the main purpose is to apply new deep learning-based clustering methods to overcome the data sparsity problem and increment the efficiency of recommender systems ...
Read More
Nowadays, with the expansion of the internet and its associated technologies, recommender systems have become increasingly common. In this work, the main purpose is to apply new deep learning-based clustering methods to overcome the data sparsity problem and increment the efficiency of recommender systems based on precision, accuracy, F-measure, and recall. Within the suggested model of this research, the hidden biases and input weights values of the extreme learning machine algorithm are produced by the Restricted Boltzmann Machine and then clustering is performed. Also, this study employs the ELM for two approaches, clustering of training data and determine the clusters of test data. The results of the proposed method evaluated in two prediction methods by employing average and Pearson Correlation Coefficient in the MovieLens dataset. Considering the outcomes, it can be clearly said that the suggested method can overcome the problem of data sparsity and achieve higher performance in recommender systems. The results of evaluation of the proposed approach indicate a higher rate of all evaluation metrics while using the average method results in rates of precision, accuracy, recall, and F-Measure come to 80.49, 83.20, 67.84 and 73.62 respectively.
Zahra Asghari Varzaneh; Soodeh Hosseini
Abstract
This paper proposed a fuzzy expert system for diagnosing diabetes. In the proposed method, at first, the fuzzy rules are generated based on the Pima Indians Diabetes Database (PIDD) and then the fuzzy membership functions are tuned using the Harris Hawks optimization (HHO). The experimental data set, ...
Read More
This paper proposed a fuzzy expert system for diagnosing diabetes. In the proposed method, at first, the fuzzy rules are generated based on the Pima Indians Diabetes Database (PIDD) and then the fuzzy membership functions are tuned using the Harris Hawks optimization (HHO). The experimental data set, PIDD with the age group from 25-30 is initially processed and the crisp values are converted into fuzzy values in the stage of fuzzification. The improved fuzzy expert system increases the classification accuracy which outperforms several famous methods for diabetes disease diagnosis. The HHO algorithm is applied to tune fuzzy membership functions to determine the best range for fuzzy membership functions and increase the accuracy of fuzzy rule classification. The experimental results in terms of accuracy, sensitivity, and specificity prove that the proposed expert system has a higher ability than other data mining models in diagnosing diabetes.
M. Saffarian; V. Babaiyan; K. Namakin; F. Taheri; T. Kazemi
Abstract
Today, Metabolic Syndrome in the age group of children and adolescents has become a global concern. In this paper, a data mining model is used to determine a continuous Metabolic Syndrome (cMetS) score using Linear Discriminate Analysis (cMetS-LDA). The decision tree model is used to specify the calculated ...
Read More
Today, Metabolic Syndrome in the age group of children and adolescents has become a global concern. In this paper, a data mining model is used to determine a continuous Metabolic Syndrome (cMetS) score using Linear Discriminate Analysis (cMetS-LDA). The decision tree model is used to specify the calculated optimal cut-off point cMetS-LDA. In order to evaluate the method, multilayer perceptron neural network (NN) and Support Vector Machine (SVM) models were used and statistical significance of the results was tested with Wilcoxon signed-rank test. According to the results of this test, the proposed CART is significantly better than the NN and SVM models. The ranking results in this study showed that the most important risk factors in making cMetS-LDA were WC, SBP, HDL and TG for males and WC, TG, HDL and SBP for females. Our research results show that high TG and central obesity have the greatest impact on MetS and FBS has no effect on the final prognosis. The results also indicate that in the preliminary stages of MetS, WC, HDL and SBP are the most important influencing factors that play an important role in forecasting.
H.5. Image Processing and Computer Vision
Z. Mehrnahad; A.M. Latif; J. Zarepour Ahmadabadi
Abstract
In this paper, a novel scheme for lossless meaningful visual secret sharing using XOR properties is presented. In the first step, genetic algorithm with an appropriate proposed objective function created noisy share images. These images do not contain any information about the input secret image and ...
Read More
In this paper, a novel scheme for lossless meaningful visual secret sharing using XOR properties is presented. In the first step, genetic algorithm with an appropriate proposed objective function created noisy share images. These images do not contain any information about the input secret image and the secret image is fully recovered by stacking them together. Because of attacks on image transmission, a new approach for construction of meaningful shares by the properties of XOR is proposed. In recovery scheme, the input secret image is fully recovered by an efficient XOR operation. The proposed method is evaluated using PSNR, MSE and BCR criteria. The experimental results presents good outcome compared with other methods in both quality of share images and recovered image.
M. Molaei; D. Mohamadpur
Abstract
Performing sentiment analysis on social networks big data can be helpful for various research and business projects to take useful insights from text-oriented content. In this paper, we propose a general pre-processing framework for sentiment analysis, which is devoted to adopting FastText with Recurrent ...
Read More
Performing sentiment analysis on social networks big data can be helpful for various research and business projects to take useful insights from text-oriented content. In this paper, we propose a general pre-processing framework for sentiment analysis, which is devoted to adopting FastText with Recurrent Neural Network variants to prepare textual data efficiently. This framework consists of three different stages of data cleansing, tweets padding, word embedding’s extraction from FastText and conversion of tweets to these vectors, which implemented using DataFrame data structure in Apache Spark. Its main objective is to enhance the performance of online sentiment analysis in terms of pre-processing time and handle large scale data volume. In addition, we propose a distributed intelligent system for online social big data analytics. It is designed to store, process, and classify a huge amount of information in online. The proposed system adopts any word embedding libraries like FastText with different distributed deep learning models like LSTM or GRU. The results of the evaluations show that the proposed framework can significantly improve the performance of previous RDD-based methods in terms of processing time and data volume.
J. Hamidzadeh; M. Moradi
Abstract
Recommender systems extract unseen information for predicting the next preferences. Most of these systems use additional information such as demographic data and previous users' ratings to predict users' preferences but rarely have used sequential information. In streaming recommender systems, the emergence ...
Read More
Recommender systems extract unseen information for predicting the next preferences. Most of these systems use additional information such as demographic data and previous users' ratings to predict users' preferences but rarely have used sequential information. In streaming recommender systems, the emergence of new patterns or disappearance a pattern leads to inconsistencies. However, these changes are common issues due to the user's preferences variations on items. Recommender systems without considering inconsistencies will suffer poor performance. Thereby, the present paper is devoted to a new fuzzy rough set-based method for managing in a flexible and adaptable way. Evaluations have been conducted on twelve real-world data sets by the leave-one-out cross-validation method. The results of the experiments have been compared with the other five methods, which show the superiority of the proposed method in terms of accuracy, precision, recall.
Kh. Aghajani
Abstract
Deep-learning-based approaches have been extensively used in detecting pulmonary nodules from computer Tomography (CT) scans. In this study, an automated end-to-end framework with a convolution network (Conv-net) has been proposed to detect lung nodules from CT images. Here, boundary regression has been ...
Read More
Deep-learning-based approaches have been extensively used in detecting pulmonary nodules from computer Tomography (CT) scans. In this study, an automated end-to-end framework with a convolution network (Conv-net) has been proposed to detect lung nodules from CT images. Here, boundary regression has been performed by a direct regression method, in which the offset is predicted from a given point. The proposed framework has two outputs; a pixel-wise classification between nodule or normal and a direct regression which is used to determine the four coordinates of the nodule's bounding box. The Loss function includes two terms; one for classification and the other for regression. The performance of the proposed method is compared with YOLOv2. The evaluation has been performed using Lung-Pet-CT-DX dataset. The experimental results show that the proposed framework outperforms the YOLOv2 method. The results demonstrate that the proposed framework possesses high accuracies of nodule localization and boundary estimation.
H. Momeni; N. Mabhoot
Abstract
Interest in cloud computing has grown considerably over recent years, primarily due to scalable virtualized resources. So, cloud computing has contributed to the advancement of real-time applications such as signal processing, environment surveillance and weather forecast where time and energy considerations ...
Read More
Interest in cloud computing has grown considerably over recent years, primarily due to scalable virtualized resources. So, cloud computing has contributed to the advancement of real-time applications such as signal processing, environment surveillance and weather forecast where time and energy considerations to perform the tasks are critical. In real-time applications, missing the deadlines for the tasks will cause catastrophic consequences; thus, real-time task scheduling in cloud computing environment is an important and essential issue. Furthermore, energy-saving in cloud data center, regarding the benefits such as reduction of system operating costs and environmental protection is an important concern that is considered during recent years and is reducible with appropriate task scheduling. In this paper, we present an energy-aware task scheduling approach, namely EaRTs for real-time applications. We employ the virtualization and consolidation technique subject to minimizing the energy consumptions, improve resource utilization and meeting the deadlines of tasks. In the consolidation technique, scale up and scale down of virtualized resources could improve the performance of task execution. The proposed approach comprises four algorithms, namely Energy-aware Task Scheduling in Cloud Computing(ETC), Vertical VM Scale Up(V2S), Horizontal VM Scale up(HVS) and Physical Machine Scale Down(PSD). We present the formal model of the proposed approach using Timed Automata to prove precisely the schedulability feature and correctness of EaRTs. We show that our proposed approach is more efficient in terms of deadline hit ratio, resource utilization and energy consumption compared to other energy-aware real-time tasks scheduling algorithms.
Amin Rahmati; Foad Ghaderi
Abstract
Every facial expression involves one or more facial action units appearing on the face. Therefore, action unit recognition is commonly used to enhance facial expression detection performance. It is important to identify subtle changes in face when particular action units occur. In this paper, we propose ...
Read More
Every facial expression involves one or more facial action units appearing on the face. Therefore, action unit recognition is commonly used to enhance facial expression detection performance. It is important to identify subtle changes in face when particular action units occur. In this paper, we propose an architecture that employs local features extracted from specific regions of face while using global features taken from the whole face. To this end, we combine the SPPNet and FPN modules to architect an end-to-end network for facial action unit recognition. First, different predefined regions of face are detected. Next, the SPPNet module captures deformations in the detected regions. The SPPNet module focuses on each region separately and can not take into account possible changes in the other areas of the face. In parallel, the FPN module finds global features related to each of the facial regions. By combining the two modules, the proposed architecture is able to capture both local and global facial features and enhance the performance of action unit recognition task. Experimental results on DISFA dataset demonstrate the effectiveness of our method.
V. Torkzadeh; S. Toosizadeh
Abstract
In this study, an automatic system based on image processing methods using features based on convolutional neural networks is proposed to detect the degree of possible dipping and buckling on the sandwich panel surface by a colour camera. The proposed method, by receiving an image of the sandwich panel, ...
Read More
In this study, an automatic system based on image processing methods using features based on convolutional neural networks is proposed to detect the degree of possible dipping and buckling on the sandwich panel surface by a colour camera. The proposed method, by receiving an image of the sandwich panel, can detect the dipping and buckling of its surface with acceptable accuracy. After a panel is fully processed by the system, an image output is generated to observe the surface status of the sandwich panel so that the supervisor of the production line can better detect any potential defects at the surface of the produced panels. An accurate solution is also provided to measure the amount of available distortion (depth or height of dipping and buckling) on the sandwich panels without needing expensive and complex equipment and hardware.
Mohammad Nazari; Hossein Rahmani; Dadfar Momeni; Motahare Nasiri
Abstract
Graph representation of data can better define relationships among data components and thus provide better and richer analysis. So far, movies have been represented in graphs many times using different features for clustering, genre prediction, and even for use in recommender systems. In constructing ...
Read More
Graph representation of data can better define relationships among data components and thus provide better and richer analysis. So far, movies have been represented in graphs many times using different features for clustering, genre prediction, and even for use in recommender systems. In constructing movie graphs, little attention has been paid to their textual features such as subtitles, while they contain the entire content of the movie and there is a lot of hidden information in them. So, in this paper, we propose a method called MoGaL to construct movie graph using LDA on subtitles. In this method, each node is a movie and each edge represents the novel relationship discovered by MoGaL among two associated movies. First, we extracted the important topics of the movies using LDA on their subtitles. Then, we visualized the relationship between the movies in a graph, using the cosine similarity. Finally, we evaluated the proposed method with respect to measures genre homophily and genre entropy. MoGaL succeeded to outperforms the baseline method significantly in these measures. Accordingly, our empirical results indicate that movie subtitles could be considered a rich source of informative information for various movie analysis tasks.
M. Mohammadzadeh; H. Khosravi
Abstract
Today, video games have a special place among entertainment. In this article, we have developed an interactive video game for mobile devices. In this game, the user can control the game’s character by his face and hand gestures. Cascading classifiers along with Haar-like features and local binary ...
Read More
Today, video games have a special place among entertainment. In this article, we have developed an interactive video game for mobile devices. In this game, the user can control the game’s character by his face and hand gestures. Cascading classifiers along with Haar-like features and local binary patterns are used for hand gesture recognition and face detection. The game’s character moves according to the current hand and face state received from the frontal camera. Various ideas are used to achieve the appropriate accuracy and speed. Unity 3D and OpenCV for Unity are employed to design and implement the video game. The programming language is C#. This game is written in C# and developed for both Windows and Android operating systems. Experiments show an accuracy of 86.4% in the detection of five gestures. It also has an acceptable frame rate and can run at 11 fps and 8 fps in Windows and Android respectively.
Fatemeh Alinezhad; Kourosh Kiani; Razieh Rastgoo
Abstract
Gender recognition is an attractive research area in recent years. To make a user-friendly application for gender recognition, having an accurate, fast, and lightweight model applicable in a mobile device is necessary. Although successful results have been obtained using the Convolutional Neural Network ...
Read More
Gender recognition is an attractive research area in recent years. To make a user-friendly application for gender recognition, having an accurate, fast, and lightweight model applicable in a mobile device is necessary. Although successful results have been obtained using the Convolutional Neural Network (CNN), this model needs high computational resources that are not appropriate for mobile and embedded applications. To overcome this challenge and considering the recent advances in Deep Learning, in this paper, we propose a deep learning-based model for gender recognition in mobile devices using the lightweight CNN models. In this way, a pretrained CNN model, entitled Multi-Task Convolutional Neural Network (MTCNN), is used for face detection. Furthermore, the MobileFaceNet model is modified and trained using the Margin Distillation cost function. To boost the model performance, the Dense Block and Depthwise separable convolutions are used in the model. Results on six datasets confirm that the proposed model outperforms the MobileFaceNet model on six datasets with the relative accuracy improvements of 0.02%, 1.39%, 2.18%, 1.34%, 7.51%, 7.93% on the LFW, CPLFW, CFP-FP, VGG2-FP, UTKFace, and own data, respectively. In addition, we collected a dataset, including a total of 100’000 face images from both male and female in different age categories. Images of the women are with and without headgear.
A. Torkaman; K. Badie; A. Salajegheh; M. H. Bokaei; Seyed F. Fatemi
Abstract
Recently, network representation has attracted many research works mostly concentrating on representing of nodes in a dense low-dimensional vector. There exist some network embedding methods focusing only on the node structure and some others considering the content information within the nodes. In this ...
Read More
Recently, network representation has attracted many research works mostly concentrating on representing of nodes in a dense low-dimensional vector. There exist some network embedding methods focusing only on the node structure and some others considering the content information within the nodes. In this paper, we propose HDNR; a hybrid deep network representation model, which uses a triplet deep neural network architecture that considers both the node structure and content information for network representation. In addition, the author's writing style is also considered as a significant feature in the node content information. Inspired by the application of deep learning in natural language processing, our model utilizes a deep random walk method to exploit inter-node structures and two deep sequence prediction methods to extract nodes' content information. The embedding vectors generated in this manner were shown to have the ability of boosting each other for learning optimal node representation, detecting more informative features and ultimately a better community detection. The experimental results confirm the effectiveness of this model for network representation compared to other baseline methods.
S. Mavaddati; S. Mavaddati
Abstract
Development of an automatic system to classify the type of rice grains is an interesting research area in the scientific fields associated with modern agriculture. In recent years, different techniques are employed to identify the types of various agricultural products. Also, different color-based and ...
Read More
Development of an automatic system to classify the type of rice grains is an interesting research area in the scientific fields associated with modern agriculture. In recent years, different techniques are employed to identify the types of various agricultural products. Also, different color-based and texture-based features are used to yield the desired results in the classification procedure. This paper proposes a classification algorithm to detect different rice types by extracting features from the bulk samples. The feature space in this algorithm includes the fractal-based features of the extracted coefficients from the wavelet packet transform analysis. This feature vector is combined with other texture-based features and used to learn a model related to each rice type using the Gaussian mixture model classifier. Also, a sparse structured principal component analysis algorithm is applied to reduce the dimension of the feature vector and lead to the precise classification rate with less computational time. The results of the proposed classifier are compared with the results obtained from the other presented classification procedures in this context. The simulation results, along with a meaningful statistical test, show that the proposed algorithm based on the combinational features is able to detect precisely the type of rice grains with more than 99% accuracy. Also, the proposed algorithm can detect the rice quality for different percentages of combination with other rice grains with 99.75% average accuracy.
Maryam Khazaei; Nosratali Ashrafi-Payaman
Abstract
Nowadays, whereas the use of social networks and computer networks is increasing, the amount of associated complex data with graph structure and their applications, such as classification, clustering, link prediction, and recommender systems, has risen significantly. Because of security problems and ...
Read More
Nowadays, whereas the use of social networks and computer networks is increasing, the amount of associated complex data with graph structure and their applications, such as classification, clustering, link prediction, and recommender systems, has risen significantly. Because of security problems and societal concerns, anomaly detection is becoming a vital problem in most fields. Applications that use a heterogeneous graph, are confronted with many issues, such as different kinds of neighbors, different feature types, and differences in type and number of links. So, in this research, we employ the HetGNN model with some changes in loss functions and parameters for heterogeneous graph embedding to capture the whole graph features (structure and content) for anomaly detection, then pass it to a VAE to discover anomalous nodes based on reconstruction error. Our experiments on AMiner data set with many base-lines illustrate that our model outperforms state-of-the-arts methods in heterogeneous graphs while considering all types of attributes.
F. Jafarinejad; R. Farzbood
Abstract
Image retrieval is a basic task in many content-based image systems. Achieving high precision, while maintaining computation time is very important in relevance feedback-based image retrieval systems. This paper establishes an analogy between this and the task of image classification. Therefore, in the ...
Read More
Image retrieval is a basic task in many content-based image systems. Achieving high precision, while maintaining computation time is very important in relevance feedback-based image retrieval systems. This paper establishes an analogy between this and the task of image classification. Therefore, in the image retrieval problem, we will obtain an optimized decision surface that separates dataset images into two categories of relevant/irrelevant images corresponding to the query image. This problem will be viewed and solved as an optimization problem using particle optimization algorithm. Although the particle swarm optimization (PSO) algorithm is widely used in the field of image retrieval, no one use it for directly feature weighting. Information extracted from user feedbacks will guide particles in order to find the optimal weights of various features of images (Color-, shape- or texture-based features). Fusion of these very non-homogenous features need a feature weighting algorithm that will take place by the help of PSO algorithm. Accordingly, an innovative fitness function is proposed to evaluate each particle’s position. Experimental results on Wang dataset and Corel-10k indicate that average precision of the proposed method is higher than other semi-automatic and automatic approaches. Moreover, the proposed method suggest a reduction in the computational complexity in comparison to other PSO-based image retrieval methods.
R. Serajeh; A. Mousavinia; F. Safaei
Abstract
Classical SFM (Structure From Motion) algorithms are widely used to estimate the three-dimensional structure of a stationary scene with a moving camera. However, when there are moving objects in the scene, if the equation of the moving object is unknown, the approach fails. This paper first demonstrates ...
Read More
Classical SFM (Structure From Motion) algorithms are widely used to estimate the three-dimensional structure of a stationary scene with a moving camera. However, when there are moving objects in the scene, if the equation of the moving object is unknown, the approach fails. This paper first demonstrates that when the frame rate is high enough and the object movement is continuous in time, meaning that acceleration is limited, a simple linear model can be effectively used to estimate the motion. This theory is first mathematically proven in a closed-form expression and then optimized by a nonlinear function applicable for our problem. The algorithm is evaluated both on synthesized and real data from Hopkins dataset.