Search

Summary: Multimodal sentiment analysis is a popular research direction in the field of affective computing. It extends unimodal‐based sentiment analysis to the environment based on multimodal information exchange. Face retrieval is the most critical technology in the current multimodal sentiment analysis field. Traditional methods for face retrieval rely on large amounts of data to train the matching relationship between text and face images. This approach suffers from several issues, including high data acquisition costs, insufficient or poor quality training data, and high computational costs for training models. To address the aforementioned issues, this paper proposes a face retrieval technique that merges a large language model with a visual base model. This technique explores the use of zero samples for training, eliminating the need to collect data and train the model to achieve face retrieval. As a result, it significantly reduces the cost requirements of data and computational power. First, an accurate natural text of face description is generated using a large language model based on cue engineering based on face‐independent features. Second, the text and images are encoded using a visual base model and mapped into the same multidimensional space for matching. The experimental results indicate that this method performs well on the CelebA dataset. Extracting 31 different sets of face feature descriptions from the CelebA dataset for retrieval achieves 72% on top1 accuracy and 93% on top3 accuracy. Compared to traditional methods, the method proposed in this paper achieves considerable face retrieval accuracy without collecting text image pairs of data and without training the model. [ABSTRACT FROM AUTHOR]

At present, in the age of computers and automation of services, deep learning (DL) technology, mainly the subset of machine learning (ML) and artificial intelligence (AI), is expressively used in innumerable domains of computer vision such as data analysis, image recognition, classification, natural language processing, and many more. It has become the foremost choice of researchers as of its effectiveness in producing decent results. This paper presents detailed and analytical literature starting from the very elementary level to the recent trends of this trending technology while focusing on the most used DL model, that is, convolutional neural network and its pretrained models for image classification and object detection. It also reviews diverse existing current literature based on this. Further, a brief introduction of AI, ML, and DL has also been presented, making the foundation for the readers. As pretrained models continuously give an upper edge to DL over ML and other technologies, 23 most popular pretrained models with their architectural diagrams have also been presented. This paper aims to summarize and analyze all the concepts used to formulate DL and its models. Also, we have emphasized more on the GoogleNet models and the entire Inception modules in detail. Finally, the fascinating applications and discussion on integral components of DL have been presented. This paper will definitely draw the attention of the students and researchers working in the area of DL and its models. [ABSTRACT FROM AUTHOR]

Summary: In China, catering O2O takeout develops rapidly; thus, there is the fierce competition in the catering O2O industry. More and more catering O2O takeout platforms focus on customer experience, especially service experience in distribution link, so it is extremely necessary to evaluate customer experience of catering O2O takeout in China scientifically and reasonably. This paper establishes a customer experience evaluation index system from the perspectives of O2O takeout platform, delivery staff, and distribution process. This paper uses the BP Neural Network Model to evaluate customer experience of catering O2O takeout, uses Principal Component Analysis to extract principal components as the input of BP neural network model, and uses the model of MATLAB Neural Network Toolbox to conduct the training simulation with customer experience data. The results show that the model can evaluate customer experience of catering O2O takeout. The analysis in this paper provides a new evaluation method of customer experience for catering O2O takeout in China. [ABSTRACT FROM AUTHOR]

Summary: Globally, the water‐level fluctuations in lakes are a dynamic and complex process. The fluctuation is characterized by higher non‐linearity and stochasticity, making it quite hard to forecast for future planning. However, the advent of machine learning algorithms in recent decades provides significant improvement in forecasting such fluctuations. This work provides a systematic review of the machine learning algorithms (ML) used for characterizing lake‐water level dynamics. Among those ML algorithms, the paper reviews seven distinct as well as popular algorithms namely, neural network model, support vector machines, extreme learning machine, artificial neuro‐fuzzy inference systems, evolutionary and hybrid algorithms, and deep learning models. The sample inputs, splitting of data, performance metrics, and its comparison are discussed with future scope. Moreover, the strength and limitations of those algorithms are also examined. Finally, the review in this paper gives a newer vision for water resource planners and hydrologists for sustainable lake management. [ABSTRACT FROM AUTHOR]

Image quality assessment is to simulate subjective human visual perception and realize image quality inference automatically. Although deep neural networks have achieved great success, the majority of them do not fully consider perception characteristics. Therefore, according to the human visual scale characteristics, we proposed an image quality assessment algorithm based on multiscale and dual domains fusion. Firstly, the original image and its phase congruency respectively input into two branches, feature pyramid and channel attention mechanism are adopted to extract multiscale features. After that, bilinear pool is used to aggregate the spatial and frequency domain characteristics of the corresponding scales, and allows arbitrary scale input to ensure that the features are extracted from the inherent quality images. Finally, the single quality score is obtained through learned weights of each scale. Comparative experiments between our approach and state‐of‐the‐art are conducted on five public databases, the results demonstrate that the proposed algorithm is not only robust to different types and across database, but also sensitive to scale. [ABSTRACT FROM AUTHOR]

Following the advances in various smart devices, there are increasing numbers of users of social network services (SNS), which allows communication and information sharing in real time without limitations on distance or space. Although personal information leakage can occur through SNS, where an individual's personal details or online activities are leaked, and various financial crimes such as phishing and smishing are also possible, there are currently no countermeasures. Consequently, malicious activities are being conducted through messages toward the users who are in follow or friend relationships on SNS. Therefore, in this paper, we propose a method of assessing follow suggestions from users with less likelihood of committing malicious activities through an information-driven follow suggestion based on a categorical classification of interests using both the images and text of user posts. We ensure the objectiveness of interest categories by defining these based on DMOZ, which is established by the Open Directory Project. The images and text are learnt using a convolutional neural network, which is one of the machine learning techniques developed with a biological inspiration, and the interests are classified into categories. Users with a large number of posts are defined as certified users, and a database of certified users is established. Users with similar interests are classified, and the similarity distances between certified users and users are measured, and a follow suggestion is generated to the certified user with the most similar interest. Using the method proposed in this paper to classify the interest categories of certified users and users, precisions of 80% and 79.8% were obtained, respectively, and the overall precision was 79.93%, indicating a good classification performance overall. It is expected that the method proposed in this paper can be used to provide follow suggestions of users with less likelihood of malicious activities based on the information posted by the user. [ABSTRACT FROM AUTHOR]

In big data and machine learning, energy analytics has shown rapid development in the past decade. With this development in energy analytics, the energy data has exponentially increased well in different areas and domains. Energy computing techniques have been used in various domains, particularly in energy data to handle the data analysis process. This paper focuses on explaining the concept and evolution of various research questions formulation, data extraction, quality valuation, search strategy, study selection and reporting the results in energy‐oriented domains to evaluate energy‐computing techniques. The thermal power plant, power system and smart grid datasets were highly used out of plenty of other datasets discussed and the ANN algorithm is most commonly used to classify the dataset. Different sets of algorithms are used for data analysis, classification, converting structured data to unstructured data, regression, visualization, and forecasting. Journal publications from the year 2012 to 2020 have reviewed and adopted a systematic approach to identify the findings of all relevant research methodologies in the defined research domain. [ABSTRACT FROM AUTHOR]

Summary: Web service recommendation as an emerging topic attracts increasing attention due to its important practical significance. As the number of available Web services continues to grow, users face the challenge of searching the most suitable services that meet their specific needs. Quality of service (QoS)‐based service recommendation becomes a popular approach to address this issue. However, existing QoS‐based service recommendation methods are inability to effectively capture valuable content and structural information from services. These methods often rely solely on low‐order explicit feature intersections in QoS information, do not fully utilize the high‐order implicit feature intersections, and ignore the rich semantic information existing in service descriptions and user preferences. To address this problem, this paper proposes a Web service recommendation method via combining knowledge distillation representation and DCNMIX quality prediction. This method combines content‐based and structure‐based service classification and service prediction based on multi‐dimensional service quality information. First, it builds a service relationship network using semantic features extracted from service descriptions. Second, it designs a graph neural network knowledge distillation framework. The teacher model extracts the knowledge of the graph neural network model, and the student model learns the structure‐based and feature‐based prior knowledge of the service relationship network. Then the student model is used to learn the knowledge of the teacher model, classify Web services, and obtain service representations. Finally, based on service representations and multi‐dimensional QoS information, it exploits the DCNMIX model to learn the explicit and implicit features intersections of Web services and obtain the prediction score and ranking of Web services. The experimental results on the ProgrammableWeb dataset show that the proposed method outperforms the state‐of‐the‐art baselines in terms of Recall, F1, Logloss, and AUC_ROC. [ABSTRACT FROM AUTHOR]

Summary: The rapid advancement and growth of technology have rendered cloud computing services indispensable to our activities. Threats and intrusions have since multiplied exponentially across a range of industries. In such a scenario, the intrusion detection system, or simply the IDS, is deployed on the network to monitor and detect any attacks. The paper proposes a feed‐forward deep neural network (FFDNN) method based on deep learning methodology using a filter‐based feature selection model. The feature selection strategy aims to determine and select the most highly relevant subset of attributes from the feature importance score for training the deep learning model. Three benchmark data sets were used to assess the experiment: CIC‐IDS 2017, UNSW‐NB15, and NSL‐KDD. In order to justify the proposed technique, a comparison was done using other learning algorithms ranging from classical machine learning to ensemble learning methods that can detect various attacks. The experiments showed that the FFDNN model with reduced feature subsets gave the highest accuracy of 99.53% and 94.45% in the NSL‐KDD and UNSW‐NB15 data sets, while the ensemble‐based XGBoost model performed better in the CIC‐IDS 2017 data set. In addition, the results show that the overall accuracy, recall, and F1 score of the deep learning algorithm are generally better for all the data sets. [ABSTRACT FROM AUTHOR]

Summary: Internet data thefts, intrusions and DDoS attacks are some of the big concerns for the network security today. Detection of these anomalies, is gaining tremendous impetus with the development of machine learning and artificial intelligence. Even now researchers are shifting the base from machine learning to the deep neural architectures with auto‐feature selection capabilities. We in this paper propose multiple deep neural network architectures which can select, co‐learn and teach the gradients of the neural network by itself with no human intervention. This is what we call as meta‐learning. The models are configured in both many to one and many to many design architectures. We combine long short‐term memory (LSTM), bi‐directional long short‐term memory (BiLSTM), convolutional neural network (CNN) layers along with attention mechanism to achieve the higher accuracy values among all the available deep learning model architectures. LSTMs overcomes the vanishing and exploding gradient problem of RNN and attention mechanism mimics the human cognitive attention that screens the network flow to obtain the key features for network traffic classification. In addition, we also add multiple convolutional layers to get the key features for network traffic classification. We get the time series analysis of the traffic done for the possibility of a DDoS attack without using any feature selection techniques and without balancing the dataset. The performance analysis is done based on confusion matrix scores, that is, accuracy, false alarm rate (FAR), sensitivity, specificity, false‐positive rate (FPR), F1 score, area under curve (AUC) analysis and loss functions on well‐known public benchmark KDD Cup'99 data set. The results of our experiments reveal that our models outperform existing techniques, showing their superiority in performance. [ABSTRACT FROM AUTHOR]

Summary: Dengue virus infection is one of the major worldwide health issues and a substantial epidemic infectious human disease. More than 2 billion humans live in dengue susceptible regions with an annual infection mortality rate is about 5%–20%. At the initial stages, it is difficult to differentiate dengue fever symptoms from other similar diseases. Therefore, early diagnosis of dengue disease can help in protecting human lives by making a preventive move before it turns into an infectious disease. Although early detection is critical to reducing dengue disease deaths, accurate dengue diagnosis requires a long time due to the numerous clinical examinations. For this, several techniques have been presented. But, the accuracy of these methods is not enough. Thus, this issue necessitates the development of a new diagnostic schema. Hence, to identify the dengue‐affected blood samples effectively, Updated Chimp Optimization Algorithm‐Artificial Neural Network (UChOA‐ANN) based diagnostic model is proposed in this paper. UChOA‐ANN‐based diagnostic model has been implemented and validated for the early detection and accurate prediction of dengue disease. Here, to improve the performance of ANN, UChOA is used for selecting the weight parameters of ANN optimally. The results show that the proposed diagnostic model has achieved maximum sensitivity, specificity, and accuracy at 92%, 96%, and 94%, respectively. In addition, the results reveal that the proposed diagnostic model has outperformed other methods for detecting dengue‐affected blood samples in terms of sensitivity, specificity, accuracy, PPV, NPV, FPR, FNR, and FDR. Our proposed diagnostic model has been capable to diagnose dengue in assisting doctors in its early stage with high accuracy. [ABSTRACT FROM AUTHOR]

Summary: Among deep‐learning‐based trackers, the siamese‐based method inspires many researchers due to its effectiveness and simplicity. However, the traditional siamese tracker has not achieved satisfactory performance due to the limited representation ability and the lack of appropriate model update strategy. To cover the shortage of siamese models, we proposed a cross‐similarity‐based siamese network with three contributions. First, we introduce a novel cross similarity module into the SiameseFC framework, which could improve the matching ability of fully convolutional networks during the tracking process. Second, we propose a novel attention weighting layer to emphasize various contributions of matching scores in different positions. This adaptive attention weighting scheme makes our tracker well adapt to the appearance change caused by pose variation, partial occlusion, and so on. Third, we develop a simple yet effective model update strategy, which exploits an independent classification model to invoke the model fine‐tuning process. Experimental results on the standard tracking benchmark show that our tracker performs much better than the baseline SiameseFC method and also achieves promising results in comparisons to other state‐of‐the‐art algorithms. [ABSTRACT FROM AUTHOR]

Summary: This paper presents a hybrid feature extraction and optimized deep learning model for the effective shot boundary detection (SBD). A cross‐guided bilateral filtering technique is used for pre‐processing. A hybrid fuzzy histogram with dual tree complex wavelet transform (FH‐DTCWT) is developed for the feature extraction to extract the visual features from the pre‐processed frames of the block. Candidate segment selection process is performed to identify the non‐boundary frames from the extracted features to improve the accuracy. Continuity matrix is created using the possible transition frames to verify that the frames are in sequential order without any gap. Classification of transition types are performed by the proposed optimized deep convolutional neural network. Optimized deep learning (DCNN‐RBESO) model is the combination of deep convolutional neural network (DCNN) and Rider bald eagle search optimization (RBESO) algorithm. To update the weight of DCNN model in learning, RBESO is utilized. MATLAB tool is used to develop the proposed SBD model. The experimental analysis is carried out for the proposed model using TREC Video Retrieval Evaluation (TRECVID) and VideoSeg datasets. The proposed model outperformed in the detection of shot boundaries in terms of precision, recall, and F1‐score. [ABSTRACT FROM AUTHOR]

In mobile ad hoc network, the routing task is considered to be more complicated issue. In order to find efficient route in MANET, the positions of each stable node in the network is identified initially. The main contribution of this paper is to identify the locations of neighboring nodes in MANET for the establishment of multi path routing in diverse mobility patterns. It also handles packet scheduling to balance the load as well as data forwarding with less communication time. The proposed work explains four phases such as stable node prediction, stability determination, route exploration, and packet dissemination. At first, the stable nodes are identified using recurrent neural network along with modified seagull optimization approach. By means of Garson's pruning based recurrent neural network accompanied with modified sea gull optimization (RMSG) algorithm, the stable neighbors are chosen. The network routing is formed by interconnecting the stable nodes from the source to the destination. When a routing link failure happens, the route recovery process will be initiated. Thus, the data packets are broadcasted in multi‐paths without any intervention. The measures namely packet delivery ratio, throughput, end‐to‐end delay, routing overhead, optimal path selection, and energy consumption are utilized to evaluate the performance of proposed approach. The experimental analysis proved that the proposed approach well performed than other compared existing approaches. [ABSTRACT FROM AUTHOR]

The emergence of networking systems and quick deployment of applications cause huge increase in cybercrimes which involves various applications like phishing, hacking, and malware propagation. However, the Ransomware techniques utilize certain device which may lead to undesirable properties which might shrink the paying‐victim pool. This paper devises a new method, namely Water Moth Flame optimization (WMFO) and deep recurrent neural network (Deep RNN) for determining Ransomware. Here, Deep RNN training is done with WMFO, and is developed by combining Moth Flame optimization (MFO) and Water wave optimization (WWO). Moreover, features are mined with opcodes and by finding term frequency‐inverse document frequency (TF‐IDF) amongst individual features. Moreover, Probabilistic Principal Component Analysis (PPCA) is adapted to choose significant features. These features are adapted in Deep RNN for classification, wherein the proposed WMFO is employed to produce optimum weights. The WMFO offered enhanced performance with elevated accuracy of 95.025%, sensitivity of 95%, and specificity of 96%. [ABSTRACT FROM AUTHOR]

Big data analytics is important for identifying and analyzing different patterns, relations, and trends within a large volume of data such as crime prediction. In this paper, we propose a new model for predicting crimes using neural‐fuzzy model. For this purpose, time of occurrence and longitude and latitude of the crime scene are used. We use k‐means algorithms to cluster crime records based on their locations. The crime pattern of each cluster was learnt by its own neural‐fuzzy network, in addition using shuffled frog leaping algorithm to determine the optimal cluster radius. The accuracy and Mean Absolute Error and Root Mean Squared Error of this method were compared with: Gaussian Process Regression, Classification and Regression Tree, and Artificial Neural Network. The experiment results showed that the proposed model was the most accurate model and had the least error to predict the occurrence of crimes so can be used as an effective tool in real‐world applications. [ABSTRACT FROM AUTHOR]

Summary: This paper presents a contribution in the development of a wireless sensor network, which can be used for building, in real time, dense air pollution maps, for compact urban areas. Such system may be useful for cyclists and pedestrians moving through the city. Based on such data, they can select the route in such a way, as to avoid the most polluted areas. An important step here may be development of miniaturized and cheap intelligent sensors, capable not only of data recording and transmitting, but also of some on‐site data processing and prediction. Such sensors require a development of small and power efficient circuit, including data processing unit integrated with an artificial neural network (ANN) in a single chip. We present a prototype chip that contains main components of such sensors, which include a programmable 10‐bit analog‐to‐digital converter, a programmable clock generator, and selected blocks of the ANN. The chip is a reconfigurable device, with many testing abilities. For this reason, one of the main challenges was a fast and efficient programming and testing tool. Such tool has been developed by us and is described in this work in detail along with selected measurement results. The presented work is an extended version of our conference paper ("Novel solutions for smart cities—creating air pollution maps based on intelligent sensors"). [ABSTRACT FROM AUTHOR]

In this paper, we present a formal model and analysis of the Advanced Microcontroller Bus Architecture (AMBA) Advanced High-performance Bus (AHB). The model is given in CSP ∥ B—an integration of the process algebra CSP and the state-based formalism B. We describe the theory behind the integration of CSP and B, and present the model in this theory. Analysis is performed using the model-checker ProB. The contribution of this paper may be summarized as follows: presentation of a formal model of the AMBA AHB protocol such that it may be used for analysis of co-design systems incorporating the bus, an evaluation of the integration of CSP and B in the production of such a model, and a demonstration and evaluation of ProB in performing this analysis. Copyright © 2009 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Summary: Wireless sensor networks (WSNs) are important building blocks of the communication infrastructure in smart cities, intelligent transportation systems, Industry, Energy, and Agriculture 4.0, the Internet of Things, and other areas quickly adopting the concepts of fog and edge computing. Their cybernetic security is a major issue and efficient methods to improve their safety and reliability are required. Intrusion detection systems (IDSs) are complex systems that discover cybernetic attacks, detect malicious network traffic, and, in general, protect computer systems. Artificial neural networks are used by a variety of advanced intrusion detection systems with outstanding results. Their successful use in the specific conditions of WSNs requires efficient learning, adaptation, and inference. In this work, the acceleration of a neural intrusion detection model, developed specifically for wireless sensor networks, is proposed and studied, especially from the learning and classification accuracy and energy consumption points of view. [ABSTRACT FROM AUTHOR]

Convolutional Neural Networks (CNNs) have made a great impact on attaining state‐of‐the‐art results in image task classification. Weight initialization is one of the fundamental steps in formulating a CNN model. It determines the failure or success of the CNN model. In this paper, we conduct a research based on the mathematical background of different weight initialization strategies to determine the one with better performance. To have smooth training, we expect the activation of each layer of the CNN model follow the standard normal distribution with mean 0 and SD 1. It prevents gradients from vanishing and leads to more smooth training. However, it was obtained that even with the appropriate weight initialization technique, a regular Rectified Linear Unit (ReLU) activation function increases the activation mean value. In this paper, we address this issue by proposing weight initialization based (WIB)‐ReLU activation function. The proposed method resulted in more smooth training. Moreover, the experiments showed that WIB‐ReLU outperforms ReLU, Leaky ReLU, parametric ReLU, and exponential linear unit activation functions and results in up to 20% decrease in loss value and 5% increase in accuracy score on both Fashion‐MNIST and CIFAR‐10 databases. [ABSTRACT FROM AUTHOR]

Summary: Various deep neural network (DNN) model watermarks have been proposed by researchers to verify copyrights for deep neural networks DNN. However, most DNN watermarking methods cannot prevent attackers from stealing and using the model. Unlike many existing approaches, this paper uses a channel pruning algorithm to protect DNN models, which verifies DNN models copyrights but also prevents the illegal use of DNN models. In this work, the pruning threshold or pruning rate is used as the secret key of a DNN model. After the secret key is distributed to multiple users, they prune the DNN model with the secret key, and the pruned and fine‐tuned model is provided to the users. The users can verify ownership of the model according to the pruning accuracy and fine‐tuning accuracy. If the secret key is incorrect, the accuracy of the model after fine‐tuning will be very low, and users will be unable to use the reasoning function of the fine‐tuned model. Based on the CIFAR‐10 and CIFAR‐100 datasets, we conducted experiments on five popular DNN models. The experimental results show that we can authorize multiple users by pruning very few channels in the convolution layers of the DNN model. [ABSTRACT FROM AUTHOR]

With the recent advances achieved by deep neural networks in image processing applications, researchers have begun exploring deep learning in pansharpening and obtained remarkable results. However, the existing methods are generally limited by their weak feature representation ability, often leading to spectral distortion or spatial blur. To generate high‐quality pansharpened images, this article proposes a novel neural network for pansharpening that includes both feature extraction and excitation mechanisms to consider important features. The neural network is modified with domain knowledge in pansharpening to fully extract spectral and spatial structures, and the proposed method outperforms traditional methods. [ABSTRACT FROM AUTHOR]

Summary: Sparse neural networks are shown to give accurate predictions competitive to denser versions, while also minimizing the number of arithmetic operations performed. However current GPU hardware can only exploit structured sparsity patterns for better efficiency. We propose a framework for generating structured multilevel block sparse neural networks by using the theory of graph products. Our Ramanujan bipartite graph product (RBGP) framework uses products of Ramanujan graphs to obtain the best connectivity for a given level of sparsity. This essentially ensures that the i.) the networks has the structured block sparsity for which runtime efficient algorithms exists, ii.) the model gives high prediction accuracy, due to the better expressive power derived from the connectivity of the graph, iii.) the graph data structure has a succinct representation that can be stored efficiently in memory. We use our framework to design a specific connectivity pattern called RBGP4 which makes efficient use of the memory hierarchy available on GPU. We benchmark our approach on image classification and machine translation tasks with an edge (Jetson Nano 2GB) as well as server (V100) GPUs. When compared with commonly used sparsity patterns like unstructured and block, we obtain significant speedups while achieving the same level of accuracy. [ABSTRACT FROM AUTHOR]

The propagation of rumors on social media has been identified as a critical problem in recent years that causes social panic or economic turmoil (to some extent), thereby giving rise to the need for faster identification. With the advancements in deep learning, researches based on neural networks become popular. Most of the existing methods extensively adopt recurrent neural networks (RNNs), such as gated recurrent unit and long short‐term memory. This results in a significant degradation in the concurrency performance of the models, implying an increased consumption of time and resources. This study proposes a model with multilayer transformer encoding blocks for detecting rumors. The self‐attention mechanism in the transformer encoding blocks provides better concurrency to the proposed model and improves its performance. The proposed model performs faster executions using less or no recurrences in comparison to other models based on RNNs. Experiments on two real social datasets verified that our model could achieve significantly better results than baseline methods. The accuracy rate increased by 1.1% and 2.2% on the Weibo and PHEME datasets, respectively. In comparison to the methods that use RNNs as a feature extractor, the training duration in the proposed model was reduced by 16% and 70% in Weibo and PHEME, respectively. [ABSTRACT FROM AUTHOR]

In order to suppress the interference in local area networks, this paper presents a Wireless Local Area Networks (WLAN) interference self-optimization method based on a Self-Organizing Feature Map (SOM) neural network model. This method trains the model by using original data sets as the initial vector set and using the whole Signal to Interference plus Noise Ratio (SINR) vector generated by the change of one Wireless Access Point (AP) channel as the basic feature. After the training, the SOM neural network can quickly locate the fault AP and optimize the network according to the changes of the network environment. Simulation results reveal that the proposed scheme can efficiently locate the AP where interference happens and optimize the interference with an improved user experience. Copyright © 2016 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

A distributed system consists of a collection of autonomous heterogeneous resources that provide resource sharing and a common platform for running parallel compute-intensive applications. The different application characteristics combined with the heterogeneity and performance variations of the distributed system make it difficult to find the optimal set of needed resources. When deployed, user applications are usually handled by application domain experts or system administrators who depending on the infrastructure provide a scheduling strategy for selecting the best candidate resource over a set of available resources. However, the provided strategy is usually generic, aimed at handling a wide array of applications and does not take into consideration specific application resource requirements. As such, an intelligent method for selecting the best resources based on expert knowledge is needed. In this paper, we propose a neural network-based multi-agent resource selection technique capable of mimicking the services of an expert user. In addition, to cope with the geographical distribution of the underlying system, we employ a multi-agent coordination mechanism. The proposed neural network-based scheduling framework combined with the multi-agent intelligence is a unique approach to efficiently deal with the resource selection problem. Results run on a simulated environment show the efficiency of our proposed method. Several scheduling simulations were conducted to compare the performance of some conventional resource selection methods against the proposed agent-based neural network technique. The results obtained indicate that the agent-based approach outperformed the classical algorithms by reducing the amount of time required to search for suitable resources irrespective of the resource size. Copyright © 2016 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Summary: Many application developers have been recently interested in applying deep learning techniques to their works but have little knowledge and experience on them. This paper presents the methods for a GUI‐based modeling tool to easily build deep learning models and automatically transform them into executable program codes. For reuse of existing deep learning codes, it introduces a method that imports such a program code, extracts the deep learning model architecture, and transforms it into a graphical representation that can be modified on a graphical interface. Meanwhile, a deep learning model with many layers is difficult to be visualized on a small display screen. To handle the difficulty, it proposes a method to identify the submodules called articulable subgraphs in a deep learning model and to organize the deep learning model into a hierarchical architecture using the nesting relationships of articulable subgraphs. It introduces a method to identify frequent substractures as well as articulable subgraphs as building blocks for deep learning models for levelwise views. The hierarchical representation easily enables to build deep learning models with many layers. The GUI‐based modeling tool employing the proposed methods makes nonexpert developers easily use deep learning techniques in their practical applications. [ABSTRACT FROM AUTHOR]

Summary: Energy load forecasting plays an important role in the smart grid, which can affect the promoting energy production and consumption decision‐making processes. In this paper, the state‐of‐the‐art deep learning (DL) neural models are used in the short‐term load forecasting, including the multilayer perceptron (MLP), the convolutional neural network (CNN), and the long short‐term memory (LSTM). A novel loss function is proposed for the load forecasting, and two commonly used benchmarks are used to verify the validity of the proposed function. The simulation results show that the mean absolute percentage error (MAPE) of the proposed loss function is 19.63% lower than cross‐entropy and 2.34% lower than mean absolute error (MAE). We compared the mentioned neural networks in different aspects, and the results show that in energy load forecasting, CNN has superior performance than MLP and LSTM in terms of high accuracy and robustness to weather changes. [ABSTRACT FROM AUTHOR]

Summary: The impact crater is the most common geological unit in the lunar surface, an important object to study the geological evolution history of the Moon, and the foundational basis for the geological dating of the Moon. Thus, the identification of impact craters is of important significance. As a new field with rapid development for more than ten years, deep learning has received attention from an increasing number of researchers because of its obvious advantages compared with the superficial model in terms of both feature extraction and modeling. To address various defects of the current method for the identification and feature extraction of impact craters, including its low efficiency and limited application range, this paper proposes a type of new impact crater identification using the convolutional neural network based on illumination and shadow quick‐positioning. First, we quickly locate the impact crater using the effect of illumination on the impact crater. Second, we train the impact crater images using the convolutional neural network to perform the feature extraction. Then, we identify impact craters in images. Finally, we classify the identified impact craters based on the roughness feature and find impact craters with unsmooth bottoms. Compared with the traditional algorithm, the algorithm of this paper breaks away from the important index of border but has a high identification rate under the specific resolution rate and can classify the identified impact craters. Such impact craters with unsmooth bottoms may overflow and form from the Earth's mantle at the bottom or may be the remnant of volcanic vents since ancient times, which are of great research significance. [ABSTRACT FROM AUTHOR]

Summary: In this paper, we propose an effective model for the similarity metrics of English sentences. In the model, we first make use of word embedding and convolutional neural network (CNN) to produce a sentence vector and then leverage the information of the sentence vector pair to calculate the score of sentence similarity. Considering the case of long‐range semantic dependencies between words, we propose a novel method transforming word embeddings to construct the three‐dimensional sentence feature tensor. In addition, we incorporate the k‐max pooling into the convolutional neural network to adapt to variable lengths of input sentences. The proposed model requires no external resource such as WordNet and parse tree. Meanwhile, it consumes very little time for training. Finally, we carried out extensive simulations to evaluate the performance of our model compared with other state‐of‐the‐art works. Experimental results on SemEval 2014 task (SICK test corpus) indicated that our model can achieve a good performance in the terms of Pearson correlation coefficient, Spearman correlation coefficient, and mean squared errors. Furthermore, experimental results on Microsoft research paraphrase identification (MSRP) indicated that our model can achieve an excellent performance in the terms of F1 and Accuracy. [ABSTRACT FROM AUTHOR]

Summary: Extracting training data semantics mainly depends on manual annotation samples prepared in advance, and semantic mapping largely relies on the prior conditions in image classification. A majority of the data samples in reality are unlabeled data, which necessitates a more efficient way to extract image semantics in classification. Deep learning, as an advanced machine learning technique, has recently received significant attention from both the academia and industry. Deep learning has the potential to extract semantic features from images in an automated way to fill up the semantic gap. This paper presents 2 novel deep learning models (ie, stacked denoising auto‐encoder and convolution deep Boltzmann machine). A stacked denoising auto‐encoder builds on a stacked auto‐encoder but uses a denoising auto‐encoder to improve the learning quality. Convolution deep Boltzmann machine combines the convolutional neural network model with the deep Boltzmann machine. These 2 models are evaluated on CIFAR‐10 and STL‐10 datasets, respectively. Experimental results show that the proposed 2 models outperform the original models in terms of both precision and recall. [ABSTRACT FROM AUTHOR]

Summary: Performance prediction for executing graph applications on distributed systems is a prerequisite to improve system performance. Especially for distributed systems optimized by sacrificing the accuracy of results to improve runtime performance, performance prediction can be used to determine accuracy‐related system parameters to achieve a tradeoff between the runtime and inaccuracy of results. This article presents an approach that predicts the runtime and inaccuracy of executing graph algorithms on bulk synchronous parallel (BSP)‐based systems to optimize system parameters by using an artificial neural network (ANN). The proposed approach samples different scales of subgraphs from the input graph by maintaining that the features of subgraphs are similar to the features of the input graph. Then it executes graph algorithm on each subgraph and extracts their runtime features. An ANN‐based performance prediction model is trained off‐line based on the extracted features and is used to predict the performance of executing graph algorithm on the complete input graph. We have validated the proposed approach by conducting single‐source shortest path, connected component, and PageRank on the BSP‐based distributed systems. The experimental results demonstrate that the prediction method can effectively predict the runtime and the inaccuracy with a relative error rate under 14% and under 25%, respectively, compared with the actual performance results. [ABSTRACT FROM AUTHOR]

Summary: Steganalysis is to detect whether or not the seemly innocent image hiding message. It is an important research topic in information security. With the development of steganography technology, steganalysis becomes more and more difficult. Some steganalysis methods have been proposed to improve the performance. Most research work concentrates on special steganography information detection and the image steganography features are designed manually. Few research works concentrate on universal steganalysis methods. In this paper, as the first several attempts, a novel image steganalysis method based on deep neural network is proposed. First, image high‐frequency features are extracted with wavelet transformation method because that most image hiding message are high frequency. Second, high‐dimensional image steganography features are extracted with deep neural networks according to the high‐frequency images and informative features combination is selected with a novel feature selection method based on entropy. Then, a parallel SVM model is proposed to build the steganalysis model based on large scale training samples. At last, the efficiency of the proposed method is illustrated through analyzing a practical image steganalysis example. [ABSTRACT FROM AUTHOR]

This document is an editorial for a special issue on "Edge computing accelerated deep learning: Technologies and applications." The traditional centralized approach to implementing machine learning (ML) and deep learning (DL) applications has limitations such as high latency, large bandwidth usage, and privacy concerns. Edge computing, which integrates mobile/wireless infrastructure and cloud datacenters, has emerged as a paradigm to address these issues. The special issue aims to promote innovative technologies and applications that accelerate DL in the distributed edge computing environment. The editorial highlights five selected papers that focus on using DL and other ML methods to address challenges in edge computing-based application scenarios and designing efficient and lightweight ML and DL models for the edge computing environment. The Lead Guest Editor expresses gratitude to the authors and reviewers for their contributions. [Extracted from the article]

Task‐based execution of graph workloads allows various ordered and unordered implementations, with tasks representing dependencies between graph vertices and edges. This work explores graph algorithms in the context of ordered and unordered task‐based implementations, that trade‐off work‐efficiency with parallelism. The monotonicity of convergent graph solutions is the reason behind the trade‐off between work‐efficiency and parallelism. This trade‐off results in variable performance‐based choices within and across different machines (CPUs and GPUs), graph algorithms, implementations (ordered, relaxed, and unordered). Input graphs also augment this choice space, with this work analyzing temporally changing graphs in addition to the static graphs explored by prior works. These algorithmic and architectural choices are first explored in this work, and it is seen that different graph workload‐input combinations perform optimally on diverse architectural configurations. The resulting choice space is analyzed and this work represents it in the form of characteristic variables that correlate with each choice space. Using these characteristic variables, this work proposes analytical and neural network models to correlate these choice spaces to find the best performing implementation. The variables and the prediction models proposed in this work are also integrated with a state‐of‐the‐art performance predictor on a multiaccelerator setup, and shows geometric performance gains of 54% on a CPU, 14% on a GPU, and 31.5% in a multiaccelerator setup over baseline implementations without performance prediction. [ABSTRACT FROM AUTHOR]

Communicating Sequential Processes for Java (JCSP) is a mature library that implements CSP-derived concurrency primitives in Java. A JCSP system is a hierarchical network of autonomous processes communicating over synchronous (optionally buffered) channels, and multiway synchronising through barriers. This paper presents a significant extension to the barrier mechanism: the fast resolution of choice between any number of barrier events, channel communications (in either direction) and timeouts. Previously, and in line with all currently released libraries and languages offering the CSP concurrency model, choice was restricted to channel inputs and timeouts. The paper demonstrates an application of alting barriers and explains the mechanisms used in their implementation that enables their use as guards in a choice. It also shows how choice over channel outputs becomes possible, as a simple consequence of having choice over barriers. Finally, an efficient implementation of CSP's broadcasting channels is presented (using a phased barrier synchronisation pattern) and a mechanism for allowing choice over these is discussed. With this extended JCSP, almost all CSP-specified systems can now be directly implemented. The new library is available under LGPL open source. Copyright © 2010 PH Welch, NCC Brown, J Moores, KV Chalmers & B Sputh. [ABSTRACT FROM AUTHOR]

Enabling e-Business applications on Grid computing has been an important topic of research as the beneficial perspective of Grids lies on solving computational intensive problems by large-scale distributed resources. In this paper we present a toolkit for centralized workload prediction and access, in order to meet the requirements of execution of commercial business processes on Grid infrastructures as well as to take advantage of their cost-effective capability for industrial users. The implementation of centralized workload prediction is achieved with an artificial neural network—available in the toolkit—each time a user completes a job execution. This approach has been adopted and validated within the framework of a GRIA IST project for aspecific industrial application, namely 3D image rendering, which meets the e-Business application's standards. The accuracy of the prediction showed promising results and in combination with the centralized access capability makes e-Business applications feasible on Grids. Copyright © 2006 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Summary: Numerous work focusing on Indian Languages for automatically recognizing characters has been witnessed in literature in the last few decades. But it was observed that only a handful of them targeted the optical character recognition of the Assamese language despite the language being widely spoken and used in many official works and activities across North‐East India. The limited contribution in this field seems to be inadequate and insufficient in their ability to recognize the handwritten characters due to dissimilarity in the people's writing style. This has motivated us to device a computational model for automatic recognition of the handwritten Assamese digits belonging to 10 different classes of 0 to 9. This study employs a convolutional neural network model (DigiNet) to learn and understand the various styles of handwritten digits. Our model exercises the convolutional neural network composed of six alternative Convolution and Pooling layers and is able to attain state‐of‐the‐art performance on the Assamese handwritten digits, achieving a test accuracy of 93.02% which is a pretty good success. The proficiency of the model architecture is also tested on the MNIST and the Bangla handwritten numeral datasets. Furthermore, our proposed model is compared with the pre‐trained VGG 19 architecture, where the performance of our model was better compared to the pre‐trained model on the Assamese as well as the Bangla handwritten numerals. [ABSTRACT FROM AUTHOR]

Summary: In the last decade, artificial intelligent systems based on neural networks have gradually become primary source for clinical decision support systems (CDSS) and are being used in diverse areas of medical diagnosis, classification, and prediction. An artificial neural network (ANN) consists of a large number of processing units which performs the computation in a parallel and distributed environment. They learn the pattern from the examples provided to it and then generalize based on the concepts they have learned while training. This paper presents a review of the current status of ANN and its variants as CDSS in various medical disciplines. The work focuses and describes the methods making use of simple ANN and use of real‐time approaches based on big data using ANN in cloud computing environment for various medical applications. Critical analysis of various methods based on smart approaches indicates that feed‐forward back propagation ANN performs sufficiently better in the domain of medicines with a high degree of accuracy. [ABSTRACT FROM AUTHOR]

Heterogeneous computing, combining devices with different architectures such as CPUs and GPUs, is rising in popularity and promises increased performance combined with reduced energy consumption. OpenCL has been proposed as a standard for programming such systems and offers functional portability. However, it suffers from poor performance portability, because applications must be retuned for every new device. In this paper, we use machine learning-based auto-tuning to address this problem. Benchmarks are run on a random subset of the tuning parameter spaces, and the results are used to build a machine learning-based performance model. The model can then be used to find interesting subspaces for further search. We evaluate our method using five image processing benchmarks, with tuning parameter space sizes up to 2.3 M, using different input sizes, on several devices, including an Intel i7 4771 (Haswell) CPU, an Nvidia Tesla K40 GPU, and an AMD Radeon HD 7970 GPU. We compare different machine learning algorithms for the performance model. Our model achieves a mean relative error as low as 3.8% and is able to find solutions on average only 0.29% slower than the best configuration in some cases, evaluating less than 1.1% of the search space. The source code of our framework is available at https://github.com/acelster/ML-autotuning. [ABSTRACT FROM AUTHOR]

Online recognition of glass defects plays an important role in improving the quality of glass products. This paper presents a conjugate gradient neural network-based method, which combines phase map and the defective image for recognition of glass defects. The boundary coordinates of the connected defect region are calculated and used to extract the defect region in the defective image correspondingly. The piecewise linear gray-level transformation is designed to reduce the noise and to enhance the signal-to-noise ratio of the defective image. The second iteration segmentation based on gray range is applied to calculate the low and high thresholds, and the ternary-valued defective image is acquired. The seven features calculated by Hu invariant moment and four features extracted from the ternary-valued defective image are used as inputs of the conjugate gradient neural network to recognize the defect type. Experimental results show that the accuracy of the recognition reaches up to 93 %. Copyright © 2016 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

By the reason of the variability of light and pedestrians' appearance, it is hard for a camera to obtain a clear human figure. Person re-identification with different cameras is a difficult visual recognition task. In this paper, a novel approach called attribute learning based on distributed deep convolutional neural network model is proposed to address person re-identification task. It shows how attributes, namely the mid-level medium between classes and features, are obtained automatically and how they are employed to re-identify person with semantics when an author-topic model is used to mapping category. Besides, considering the ability to operate on raw pixel input without the need to design special features, deep convolutional neural network is employed to generate features without supervision for attributes learning model. To overcome the model's weakness in computing speed, parallelized implementations such as distributed parameter manipulation and attributes learning are employed in attribute learning based on distributed deep convolutional neural network model. Experiments show that the proposed approach achieves state-of-the-art recognition performance in the VIPeR data set and is with a good semantic explanation. Copyright © 2016 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Summary: Research on tire‐ice traction theory is still in the basic stage, and there are large differences between the current theoretical calculation and engineering practice. Because testing of pneumatic tire performance on icy road is very expensive, and testing conditions are complex and variable, testing does not have higher repeatability. In this paper, a tire‐ice traction model is established based on a neural network and the finite‐element method. Through the model calculations, the traction characteristics of tires on ice can be well‐simulated. The method of combining the Levenberg‐Marquardt (LM) optimizing algorithm with the neural network is used to predict the friction between tires and ice. The attributive characteristics of tires based on testing results are then measured. A nonlinear finite‐element model of a 37 × 12.5R16.5 tire is established by adopting the Yeoh model of rubber and researching existing tire‐ice driving traction models. This tire‐ice finite‐element model can predict the changes of drawbar pull and torque when driving on icy roads, and two different tire working conditions, ie, straight line and turning, are taken into consideration simultaneously. The influence of slip ratio, speed, load on drawbar pull, and torque are also explored. [ABSTRACT FROM AUTHOR]

Summary: Drowsy student state detection is helpful to understand the students' learning state, which is the necessary and basic aspect of teaching activities evaluation and assessment. The performance of traditional methods may deteriorate dramatically because of the external environment factors. In this paper, a novel drowsy student state detection method by integrating deep convolutional neural network is proposed at the first time in the literature. The proposed method avoids the complicated manual feature extraction operation and it can effectively reduce the interference of environmental factors in the application scenarios. Experimental results demonstrate that our approach can achieve high accuracy and lower error rate for drowsy student state detection. In addition, the results also show that our method outperforms traditional methods. [ABSTRACT FROM AUTHOR]

The precise prediction of bus routes or the arrival time of buses for a traveler can enhance the quality of bus service. However, many social factors influence people's preferences for taking buses. These social factors may include heavy traffic cost, traffic congestion, poor air quality and so forth. Existing prediction techniques rarely consider social sensing when predicting the bus arrival time. Accordingly, this paper proposes a social sensing enhanced service for predicting bus routes, which integrates sensing ability and social networks to understand and measure the influence between social events and vehicle velocity. We focus on the analysis of two different attributions: PT service quality attributions PEAs and road condition attributions PRCAs. Both of them synthesize the social sensing in their evaluation of bus routes. PEA represents individual preferences and PRCA represents physical factors that significantly influence vehicle velocity. Bus relevant social events were further categorized into PEA events or PRCA events. PEAs of buses were scored according to the tendency of bus conditions reflected in social events. Furthermore, an artificial neural network prediction model is established to estimate the bus travel time. Copyright © 2015 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Summary: Cerebral micro‐bleedings are small chronic brain hemorrhages caused by structural abnormalities of the small vessels. CMBs can be found from individuals with stroke at memory clinics and even healthy elderly people. CMBs indicate hemorrhage‐prone pathological states. Research shows that CMBs are associated with an increased risk of future ischemic stroke, intra‐cerebral hemorrhage (ICH), dementia, and death. Considering that CMBs severely influence people's life, it is necessary to identify the CMBs in an early stage to prevent from further deterioration and to help people live a healthy life. In this paper, we proposed using CNN with stochastic pooling for the CMB detection. CNN has good performance in image and video recognition, recommender system, and nature language processing. Based on the collected subject, the experiment result shows that the six‐convolution layer and three fully‐connected layer CNN, nine‐layers in total, achieved sensitivity, specificity, accuracy, and precision as 97.22%, and 97.35%, 97.28%, and 97.35% in average of ten runs, which shows better performance than five state‐of‐the‐art methods. [ABSTRACT FROM AUTHOR]

Summary: Doctors always desire high‐resolution medical images to have accurate diagnosis. Super‐resolution (SR) is a technology that can improve the resolution of medical images. Convolutional neural network (CNN)–based SR methods have achieved desired performance in natural images. In this paper, we apply a deep dense SR (DDSR) convolutional neural networks model to two types of medical images, including Computerized Tomography (CT) images and Magnetic Resonance imaging (MRI) images. This network densely connects every hidden layer to learn high‐level features, which was first proposed for object recognition. A set of medical images is used for experiments. We compare the performance of DDSR with three state‐of‐the‐art SR network models, including SR Convolutional Neural Network (SRCNN), Fast SR Convolutional Neural Network (FSRCNN), and Very Deep SR Convolutional Neural Network (VDSR). Both the objective indices and subjective evaluations are used for comparison. The results show that the proposed network has better performances both on CT and MRI images. [ABSTRACT FROM AUTHOR]

Summary: Artificial intelligence is widely believed to reshape many industries in the future. This paper analyzes the chaotic characteristics of Chinese export price index, exchange rate data, and studies the predictive ability of the neural network model in exchange rate passthrough (ERPT) prediction. The Hermite neural network model, one of artificial intelligence models, is selected to forecast the rate of pass‐through effect. Hermite neural network model demonstrates a superior forecasting accuracy, and its forecasting error is less than that of the general neural network model. [ABSTRACT FROM AUTHOR]

Summary: For improving the speed of character recognition, this paper applies data mining technology to character recognition under the condition of big data architecture. As an effective method of data mining, artificial neural network is a computational model established by simplifying, abstracting, and simulating the biological neural mechanism of human brain. Artificial neural network ensemble method, by training several weak learning devices to form the final strong learning device, can better improve the learning generalization ability of the model, and it has a broad application prospect. The main difficulties in the research of license plate recognition are the low quality of vehicle license plate image, the small proportion of license plate area, the distortion of characters to be recognized, and the large interference noise. In this paper, wavelet transform, edge detection, and line scanning methods are used to realize multi‐size and multi‐type license plate location, which overcomes the influence of illumination, color, size, and position. In the character recognition module, the improved BP neural network and the neural network integrated classifier based on weight clipping are used to identify. Experimental results show that the proposed data mining algorithm can further improve the learning efficiency and improve the performance of the time. [ABSTRACT FROM AUTHOR]

Summary: Deep learning has emerged as an effective solution to various text mining problems such as document classification and clustering, document summarization, web mining, and sentiment analysis. In this paper, we describe our work on investigating several deep learning models for a binary sentiment classification problem. We used movie reviews in Turkish from the website www.beyazperde.com to train and test the deep learning models. We also report a detailed comparison of the models in terms of accuracy and time performances. Two major deep learning architectures used in this study are Convolutional Neural Networks and Long Short‐Term Memory. We built several variants of these models by changing the number of layers, tuning the hyper‐parameters, and combining models. Additionally, word embeddings were created by applying the word2vec algorithm with a skip‐gram model on a large dataset (∼ 13 M words) composed of movie reviews. We investigate the effect of using the pre‐word embeddings with these models. Experimental results have shown that the use of word embeddings with deep neural networks effectively yields performance improvements in terms of run time and accuracy. [ABSTRACT FROM AUTHOR]