In service computing, combining multiple services through service composition to address complex user requirements has become a popular research topic. QoS-aware service composition aims to find the optimal composition scheme with the QoS attributes that best match user requirements. However, certain QoS attributes may continuously change in a dynamic service environment, so service composition methods need to be adaptive. Furthermore, the large number of candidate services poses a key challenge for service composition, where existing service composition approaches based on reinforcement learning (RL) suffer from low efficiency. To deal with the problems above, in this paper, a new service composition approach is proposed which combines RL with skyline computing where the latter is used for reducing the search space and computational complexity. A WSC-MDP model is proposed to solve the large-scale service composition within a dynamically changing environment. To verify the proposed method, a series of comparative experiments are conducted, and the experimental results demonstrate the effectiveness, scalability and adaptability of the proposed approach.

Threshold secret sharing is a fundamental building block in information security to provide secrecy and robustness services for various cryptographic protocols. According to the definition of (t, n) threshold secret sharing, the secret is divided into n shares, such that any t or more than t of these shares allow the secret to be reconstructed; but less than t shares reveal no information of the secret. In other words, this definition only considers protection of the secret from colluded insiders but not outsiders. In this paper, we propose an extended secret sharing scheme, called secret sharing with secure secret reconstruction, in which the secret can be protected in the reconstruction phase from both attacks of insiders and outsiders. In traditional secret sharing schemes, when more than t shares are presented in the secret reconstruction, outsiders only need to intercept t shares to recover the secret. But in our proposed basic scheme, outsiders need to intercept all the released shares to recover the secret. Obviously, requiring more shares in the reconstruction contributes to security enhancement for this process. The limitation of this basic scheme is that it cannot prevent outsiders from learning the secret if they intercept all the released shares. To address this issue, we further extend the basic scheme so that the reconstructed secret is only accessible to shareholders, but not to outsiders. To the best of our knowledge, our extended scheme is the first secret sharing scheme that satisfies this property with information theoretical security.

Inversion of neural networks aims to find optimal input variables given a target output, and is widely applicable in an industrial field such as optimizing control variables of complex systems in manufacturing facilities. To achieve optimal inputs using a standard first-order optimization technique, proper initialization of input variables is essential. This paper presents a new initialization method for input variables of neural networks based on k-nearest neighbor (k-NN) approach. The proposed method finds inputs which resulted in an output close to a target output in a training dataset, and combine them to form initial input variables. Experiments on a toy dataset demonstrate that our method outperforms random initialization. Also, we introduce an exhaustive case study on power scheduling of a heating, ventilation, and air conditioning (HVAC) system in a building to support the effectiveness of the algorithm.

Learning from imbalanced datasets poses a major challenge in data mining community. When dealing with imbalanced datasets, conventional classification algorithms generally perform poorly as they are originally designed to work under balanced class distribution scenarios. Although there exist different methods to addressing this issue, sampling methods especially over-sampling techniques have shown great potentials as they aim to improve datasets itself rather than the classifiers, which can allow them to be used for any classifier. In this paper, we propose a novel adaptive weighted over-sampling for imbalanced datasets based on density peaks clustering with heuristic filtering. Unlike other clustering-based over-sampling methods, the proposed approach applies modified density peaks clustering rather than traditional k-means clustering techniques to cluster the minority instances due to its capability of accurately identifying sub-clusters with different sizes and densities, which is beneficial for the proposed method to simultaneously accommodate for between-class and within-class imbalance issues caused by various reasons. Subsequently, the size for each identified sub-cluster to be oversampled is adaptively determined according to its own size and density and then the minority instances within each sub-cluster are oversampled based on their probabilities inversely proportional to their distances to the majority class and their densities with the aim of generating more synthetic minority instances for borderline and sparser ones. Finally, in order to avoid the generation of overlapping, a heuristic filtering strategy is also developed to iteratively move the possibly overlapped minority instances away from the majority class. The extensive experimental results on the different imbalanced datasets demonstrate that the proposed approach can achieve better classification performance in most datasets as compared to the other existing over-sampling techniques.

Accurate and real-time traffic passenger flows forecasting at transportation hubs, such as subway/bus stations, is a practical application and of great significance for urban traffic planning, control, guidance, etc. Recently deep learning based methods are promised to learn the spatial-temporal features from high non-linearity and complexity of traffic flows. However, it is still very challenging to handle so much complex factors including the urban transportation network topological structures and the laws of traffic flows with spatial and temporal dependencies. Considering both the static hybrid urban transportation network structures and dynamic spatial-temporal relationships among stations from historical traffic passenger flows, a more effective and fine-grained spatial-temporal features learning framework is necessary. In this paper, we propose a novel spatial-temporal incidence dynamic graph neural networks framework for urban traffic passenger flows prediction. We first model dynamic traffic station relationships over time as spatial-temporal incidence dynamic graph structures based on historically traffic passenger flows. Then we design a novel dynamic graph recurrent convolutional neural network, namely Dynamic-GRCNN, to learn the spatial-temporal features representation for urban transportation network topological structures and transportation hubs. To fully utilize the historical passenger flows, we sample the short-term, medium-term and long-term historical traffic data in training, which can capture the periodicity and trend of the traffic passenger flows at different stations. We conduct extensive experiments on different types of traffic passenger flows datasets including subway, taxi and bus flows in Beijing. The results show that the proposed Dynamic-GRCNN effectively captures comprehensive spatial-temporal correlations significantly and outperforms both traditional and deep learning based urban traffic passenger flows prediction methods.

Zero-shot learning (ZSL) in visual classification aims to recognize novel categories for which few or even no training samples are available. Through recent advances using generative adversarial networks (GANs) for cross-modal generation, several generative methods have been investigated for ZSL to classify unseen categories with synthetic samples. However, these GAN-based ZSL approaches still struggle to generate samples with semantic consistency and significant between-class discrepancy while preserving within-class diversity, which are vital to building classifiers for unseen classes. Accordingly, in this paper, we propose a robust dual-stream GAN to synthesize satisfactory samples for zero-shot visual classification. In more detail, the inter-class discrepancy is maximized by a backbone compatibility loss, which drives the center of the synthesized samples to move towards the center of real samples of the same class while moving further away from samples of different classes. Secondly, in order to preserve the intra-class diversity ignored by most extant paradigms, we propose a stochastic dispersion regularization to encourage the synthesized samples to be distributed at arbitrary points in the visual space of their categories. Finally, unlike previous methods that project visual samples back into semantic space and consequently cause an information degradation problem, we design a dual-stream generator to synthesize visual samples and reconstruct semantic embedding simultaneously, thereby ensuring semantic consistency. Our model outperforms the state-of-the-arts by 4.7% and 3.0% on average in two metrics over four real-world datasets, demonstrating its effectiveness and superiority.

In this article, we prove that the generalized difference between A,B∈RFC, i.e., fuzzy numbers with continuous endpoints, is given by an interactive difference. To be more precise, we construct a certain joint possibility distribution I such that the generalized difference coincides with the sup-I extension of the subtraction. As an immediate consequence, we have that every notion of difference between A,B∈RFC, that has so far appeared in the literature, can be derived from a sup-J extension for some particular choice of J. Moreover, we show that both the generalized and the generalized Hukuhara derivative of a function f:R→RFC at x∈R can be expressed as the limit for h → 0 of a difference quotient, where the difference is an interactive difference for each h. For short, we say that the generalized (as well as the generalized Hukuhara) difference is interactive.

In three-way decision, the description on the risk attitude of decision-makers is a focus topic. In this paper, we propose a novel three-way decision model based on cumulative prospect theory. First, with the aid of a reference point, the value functions are utilized to describe different risk appetites of decision-makers towards gains and losses. Second, the weight functions incorporate the nonlinear transformation of the conditional probability. The cumulative decision weights are decided by taking into account the increasing order of value functions. Then, with value functions and weight functions, the new decision rules of the proposed model are deduced based on the principle of maximizing the cumulative prospect value rather than minimizing the cost. Further, we analyze and prove the existence and uniqueness of thresholds of our model. Then, the decision rules are simplified based on the conditional probability and numerical solutions of thresholds, and the algorithm for deriving three-way decision rules is constructed. Finally, an illustrative example and a series of relevant comparisons are presented to illustrate and validate the effectiveness and feasibility of our model.

Information propagation modeling and influence maximization are two important research problems in viral marketing. When marketing information is given, how can the seed nodes be efficiently identified to maximize the spread of the information through the network? To answer this question, we consider multiple factors in information propagation, such as information content, social influence and user authority, and propose a multifactor-based information propagation model (MFIP). Then, we utilize the first-order influence of the nodes to approximate their influence and propose an efficient heuristic algorithm named weighted degree decrease (WDD) to select the seed nodes under the MFIP model. Experimental evaluations with four real-world social network datasets demonstrate the effectiveness and efficiency of our algorithm.

This paper investigates the synchronization of multiple memristive neural networks (MMNNs) under cyber-physical attacks through distributed event-triggered control. In the field of multi-agent dynamics, memristive neural network (MNN) is considered as a kind of switched systems because of its state-dependent parameters which can lead to the parameters mismatch during synchronization. This will increase the uncertainty of the system and affect the theoretical analysis. Also, neural network is considered as a typical nonlinear system. Therefore, the model studied in this paper is a nonlinear system with switching characteristics. In complex environments, MMNNs may receive mixed attacks, one of which is called cyber-physical attacks that may influence both communication links and MNN nodes to cause changes in topology and physical state. To tackle this issue, we construct a novel Lyapunov functional and use properties of M-matrix to get the criteria for synchronization of MMNNs under cyber-physical attacks. It is worth mentioning that the controllers in this paper are designed to be distributed under event-triggering conditions and Zeno behavior is also excluded. In addition, the algorithm of parameter selection is given to help designing the controllers. One example is given at the end of the paper to support our results.

The possibility distribution-based approach is one of the powerful tools available to manage hesitant fuzzy linguistic term set (HFLTS) information. However, existing possibility distribution studies have not considered the experts’ satisfied preference for HFLTSs in the process of generating the possibility distribution. This paper aims at filling this research gap. To achieve this goal, a novel possibility distribution generation method based on the concept of linguistic quantifier is proposed. This is accomplished by defining a new attitude linguistic quantifier, which is supported with theoretical results to analyze the relationship between the proposed attitude linguistic quantifier with the original linguistic quantifier, attitude indices and the expected linguistic term. The new possibility distribution generation method is proved to be (1) more general than the two main existing approaches, which are particular cases for specific linguistic quantifiers; and (2) useful to implement the concept of soft majority in the resolution process of the decision making situation. Additionally, a new two stages feedback mechanism of attitude adjustment and assessment adjustment is devised to guarantee the convergence of the consensus reaching process. Finally, a framework of group decision making with HFLTSs information is presented and an illustrative example is conducted to verify the proposed method.

The Latent Dirichlet Allocation (LDA) model, which is a document-level probabilistic model, has been widely used topic modeling. However, an essential issue of the model is the shortage in identifying co-occurrence relationships (e.g., aspect-aspect, aspect-opinion, etc.) in sentences. To address the problem, we propose an association constrained LDA (AC-LDA) for effectively capturing the co-occurrence relationships. Specifically, based on the basic features of the syntactic structure in product reviews, we then formalize three major types of word association combinations and carefully design corresponding identification. For reducing the influence of global aspect words on the local distribution, we apply an important constraint on global aspects. Finally, the constraint and related association combinations are merged into the LDA to guide the topic-words allocation in the learning process. Based on the experiments on real-world product review data, we demonstrate that our model can effectively capture the relationship hidden in local sentences and further increase the extraction rate of fine-grained aspects and opinion words. Our results confirm the superiority of the AC-LDA over the state-of-the-art technique in terms of the extraction accuracy. We also verify the strength of our method in identifying irregularly appeared terms, such as non-aspect opinions, low-frequency words, and secondary aspects.

This paper focuses on a twofold relational generalization of the notion of Galois connection. It is twofold because it is defined between sets endowed with arbitrary transitive relations and, moreover, both components of the connection are relations, not necessarily functions. A characterization theorem of the notion of relational Galois connection is provided and, then, it is proved that a suitable notion of closure can be obtained within this framework. Finally, we state a necessary and sufficient condition that allows to build a relational Galois connection starting from a single transitive digraph and a single binary relation.

Immunizing a subset of nodes in a network - enabling them to identify and withstand the spread of harmful content - is one of the most effective ways to counter the spread of malicious content. It has applications in network security, public health policy, and social media surveillance. Finding a subset of nodes whose immunization results in the least vulnerability of the network is a computationally challenging task. In this work, we establish a relationship between a widely used network vulnerability measure and the combinatorial properties of networks. Using this relationship and graph summarization techniques, we propose an efficient approximation algorithm to find a set of nodes to immunize. We provide theoretical justifications for the proposed solution and analytical bounds on the runtime of our algorithm. We empirically demonstrate on various real-world networks that the performance of our algorithm is an order of magnitude better than the state of the art solution. We also show that in practice the runtime of our algorithm is significantly lower than that of the best-known solution.

Network embedding has recently attracted lots of attention due to its wide applications on graph tasks such as link prediction, network reconstruction, node stabilization, and community stabilization, which aims to learn the low-dimensional representations of nodes with essential features. Most existing network embedding methods mainly focus on static or continuous evolution patterns of microscopic node and link structures in networks, while neglecting the dynamics of macroscopic community structures. In this paper, we propose a Community-aware Dynamic Network Embedding method (short for CDNE) which considers the dynamics of macroscopic community structures. First, we model the problem of dynamic network embedding as a minimization of an overall loss function, which tries to maximally preserve the global node structures, local link structures, and continuous community dynamics. Then, we adopt a stacked deep autoencoder algorithm to solve this minimization problem, obtaining the low-dimensional representations of nodes. Extensive experiments on both synthetic networks and real networks demonstrate the superiority of CDNE over the existing methods on tackling various graph tasks.

Although most of the control design methods assume unbounded control signals, real systems do have saturating actuators, which may degenerate closed-loop performance or even lead to unstable behavior. Additionally, the delay is generally almost ubiquitous in processes, that is also imposing performance and stability constraints. Our main contribution is to provide a controller design methodology for the stabilization of delayed systems under saturating actuators. Specifically, we address the design of non-Parallel Distributed Compensation (non-PDC) state feedback fuzzy control laws that locally stabilize a class of nonlinear discrete-time systems with state time-varying delay and saturating actuators. The proposed non-PDC control law depends on the current state xk and the state delayed by d¯ samples. Based on the Lyapunov-Krasovskii approach, we characterize the safe region of initial conditions through two sets: an ellipsoidal one for the current state vector, and another set for the delayed state vectors. Trough two convex optimization procedures, we can maximize the estimate of the region of attraction of the closed-loop control system. Additionally, a relaxation method inspired by the Frank-Wolfe algorithm is introduced, yielding better estimates of the region of attraction. The achievements are compared with other finds in the literature, illustrating the efficiency of this proposal.

Graphs and networks are very common data structure for modelling complex systems that are composed of a number of nodes and topologies, such as social networks, citation networks, biological protein-protein interactions networks, etc. In recent years, machine learning has become an efficient technique to obtain representation of graph for downstream graph analysis tasks, including node classification, link prediction, and community detection. Different with traditional graph analytical models, the representation learning on graph tries to learn low dimensional embeddings by means of machine learning models that could be trained in supervised, unsupervised or semi-supervised manners. Compared with traditional approaches that directly use input node attributes, these embeddings are much more informative and helpful for graph analysis. There are a number of developed models in this respect, that are different in the ways of measuring similarity of vertexes in both original space and feature space. In order to learn more efficient node representation with better generalization property, we propose a task-independent graph representation model, called as graph deconvolutional network (GDN), and corresponding unsupervised learning algorithm in this paper. Different with graph convolution network (GCN) from the scratch, which produces embeddings by convolving input attribute vectors with learned filters, the embeddings of the proposed GDN model are desired to be convolved with filters so that reconstruct the input node attribute vectors as far as possible. The embeddings and filters are alternatively optimized in the learning procedure. The correctness of the proposed GDN model is verified by multiple tasks over several datasets. The experimental results show that the GDN model outperforms existing alternatives with a big margin.

In real life, there are many complex multiple attribute group decision making (MAGDM) problems with high decision risk and uncertainty. The decision-making process of the complex MAGDM can encounter the following three problems: (1) Since different experts have different knowledge structures and interests, they master different individual attribute information of alternatives. (2) Experts may have different consensus degrees for alternatives under different attributes. (3) For some alternatives, the experts can not make an immediate decision in the actual decision-making process. The experts need much more information to decide on these alternatives in the subsequent decision step. To solve the problems as mentioned above, we propose sequential three-way multiple attribute group decision making (STWMAGDM) with individual attributes by introducing sequential three-way decisions. Meantime, we construct a multilevel granular structure based on the consensus degree of attributes. Further, at each granularity level, the experts need to reach consensus before deducing decision results. For improving the consensus reaching process, we take into account the social influence among experts with the aid of opinion dynamics. In this case, we construct social networks based on the similarity of experts and the amount of attribute information mastered by experts to describe the social influence. Meanwhile, we modify the model of opinion dynamics by introducing the interaction willingness of experts and establish the corresponding adjustment rules of interaction willingness. Finally, we use two diagnosis examples of breast cancer and heart disease to explain our model in detail. In order to verify the effectiveness of our method, we also perform the corresponding comparative experiments and sensitivity analyses.

Inverse reinforcement learning (IRL) involves imitating expert behaviors by recovering reward functions from demonstrations. This study proposes a model-free IRL algorithm to solve the dilemma of predicting the unknown reward function. The proposed end-to-end model comprises a dual structure of autoencoders in parallel. The model uses a state encoding method to reduce the computational complexity for high-dimensional environments and utilizes an Adaboost classifier to determine the difference between the predicted and demonstrated reward functions. Relative entropy is used as a metric to measure the difference between the demonstrated and the imitated behavior. The simulation experiments demonstrate the effectiveness of the proposed method in terms of the number of iterations that are required for the estimation.

The key aspect in coal production is realizing safe and efficient mining to maximize the utilization of the resources. A requirement for sustainable economic development is realizing green coal production, which is influenced by factors of coal economic, energy, ecological, coal gangue economic and social benefits. To balance these factors, this paper proposes a many-objective optimization model with five objectives for green coal production. Furthermore, a hybrid many-objective particle swarm optimization (HMaPSO) algorithm is designed to solve the established model. A new offspring of the alternative pool is generated by employing different evolutionary operators. The environmental selection mechanism is adopted to select and store the excellent solutions. Two sets of experiments are performed to verify the effectiveness of the proposed approach: First, the HMaPSO algorithm is tested on the DTLZ functions, and its performance is compared with that of several widely used many-objective algorithms. Second, the HMaPSO algorithm is applied to solve the many-objective green coal production optimization model. The computational results demonstrate the effectiveness of the proposed approach, and the simulation results prove that the designed approach can provide promising choices for decision makers in regional planning.

To generate solutions to multiple criteria group decision-making (MCGDM) problems that are satisfactory to the decision makers, this paper proposes a new method. To examine whether a group solution is satisfactory to the decision makers, group satisfaction is constructed from alternative assessment and ranking differences between the decision makers and the group. The difference between a decision maker's assessment and a group's assessment is designed based on differences in assessment grades, whose normalization is theoretically proven to construct alternative assessment differences. Inspired by Spearman's rank correlation coefficient, the expected utilities of decision makers’ and the group's assessments are used to construct alternative ranking differences. An abstract two-variable function with specific properties is designed to relate alternative assessment difference to alternative ranking difference to form group satisfaction. From the constructed group satisfaction, the process of generating group-satisfactory solutions to MCGDM problems is presented. The problem of selecting engineering project management software is analyzed by using the proposed method to demonstrate its applicability. To highlight the importance of group satisfaction in MCGDM, relationships and differences between group satisfaction and group consensus are analyzed through the problem and simulation experiments.

This study systematically investigated the use of singular value decomposition (SVD) for the blind watermarking of color images. The proposed algorithm overcomes most of the problems typically encountered when using existing SVD-based schemes, while concurrently enhancing performance in terms of imperceptibility and robustness. After applying SVD to non-overlapping 4 × 4 image blocks, level shifting is used to control the embedding strength in accordance with the intensity of pixels in each block. The proposed watermark embedding process helps to preserve orthonormality in the unitary matrix and compensate for the resulting distortion. Iterative regulation ensures the accurate retrieval of the embedded watermark, while mixed modulation helps to improve robustness without compromising image quality. Experiment results demonstrate that the proposed watermarking algorithm is highly resistant to a variety of image processing attacks and error-free in the absence of attack. The proposed method outperforms existing SVD-based schemes in terms of imperceptibility and robustness at a payload capacity of 1/16 bit per pixel.

Chaos-based encryption provides a practical way to protect the confidentiality of digital images nowadays. The increasing convenience (e.g., larger bandwidth) of data sharing stimulates the need for encrypting amounts of images in a fast manner. Yet most existing works aim to encrypt an image for each time. Although some parallel encryptions have been proposed, the speed is still far from satisfactory to proceed with the huge increasing number of images. This inspires us to consider another promising way, encrypting a batch of images parallelly for each time. We use maximum available number of threads in parallel computation for full use of processor resources. Considering the batch images as a shared resource, every thread competes with others to encrypt images in the shared resource in a preemptive manner for encryption. A classical permutation-diffusion architecture for chaos-based encryption is utilized for each thread, where logistic map and Lorenz system are used for generating keystream for permutation and diffusion, respectively. We make cryptographical analyses and perform experiments to confirm that the security is guaranteed. The results of efficiency tests demonstrate that the encryption speed is greatly improved compared with the state-of-art image encryption algorithms in parallel as well as serial modes.

This paper proposes an event-based near-optimal control algorithm for nonaffine discrete-time systems with constrained inputs. The method is derived from the dual heuristic dynamic programming (DHP) technique. The challenge caused by saturating actuators is overcome by using a nonquadratic performance index. Then, the event-based control technique is used to decrease the amount of computation. Meanwhile, the stability analysis is provided. It illustrates that the proposed event-based method can asymptotically stabilize the nonaffine systems by using the Lyapunov method. Furthermore, the stability conditions and the design process of the event-based controller are established. The event-based DHP algorithm is implemented by constructing three neural networks, namely, the model network, the critic network, and the action network. Finally, simulation studies are conducted to demonstrate the applicability and the performance of the proposed method.

Correlation-filter-based trackers have shown favorable accuracy and efficiency in visual tracking. However, most of these trackers are prone to drift in cases of heavy occlusions and temporal tracking failures because they only maintain the short-term memory of target appearance via a highly adaptive update mode. In this paper, we propose a reliable visual tracking method based on a dual-memory selection (DMS) model to alleviate tracking drift. Considering that long-term memory is robust to heavy occlusions while short-term memory performs well in rapid appearance changes, the proposed DMS model combines these two memory patterns of the target appearance and adaptively selects a reliable memory pattern to handle the current tracking challenges via a memory selector. For each memory pattern, a memory tracker is established based on discriminative correlation filters. The short-term tracker aggressively updates the target model to capture recent appearance changes via a linear interpolation update model, while the long-term tracker conservatively updates the target model to maintain historical appearance characteristics with a memory-improved update model and a dynamic learning rate. Furthermore, a novel memory evaluation criterion (MEC) is developed to evaluate the reliability of each tracker for memory selection. From credibility and discriminability measurements considering the temporal context, the memory tracker with the highest reliability score is selected to determine the target location in each frame. Extensive experiments on public benchmark datasets demonstrate that the proposed tracking method performs favorably compared to multiple recent state-of-the-art methods.

In this paper, we propose a simple and effective method for creating accurate trimaps based on input images. Most advanced matting algorithms require the user to provide prior information to estimate high-quality alpha masks, where the prior information is primarily in the form of trimaps. A precise trimap is one of the most important factors affecting the performance of the matting algorithm. It is a very tedious task for users to specify a large number of accurate trimaps, and it is even impractical in some applications. Based on manifold ranking, we use strokes to mark the superpixel nodes to create high-quality trimaps. The experimental results show that the method given in this paper can generate high-quality trimaps, thus ensuring the accuracy of the alpha masks that are estimated by the matting algorithm. We verify the performance of the trimaps that are created using the method given in this paper for various matting algorithms.

One-shot semantic segmentation aims to recognize unseen object regions by using the reference of only one annotated example. Many deep convolutional neural networks have achieved enormous success on this task. However, most of the existing methods only use a fixed annotated dataset to train the network. The remaining unannotated examples remain difficult to be leveraged and recognized. In this study, we propose a procedure based on the generative adversarial network to enable the one-shot semantic segmentation model for learning information from previously unknown categories. Our method contains a segmentation network that generates segmentation predictions. We then use a discriminator to differentiate the probability maps of segmentation prediction from the ground truth distribution. Consequently, we can ignore the pixels classified as fake and only use trustworthy regions as the label to train the segmentation network, thus achieving semi-supervised learning. Experimental results demonstrate the effectiveness of the proposed adversarial learning method with an average gain of 49.7% accuracy score on the PASCAL VOC 2012 dataset.

A number of approaches to graph-based subspace clustering, which assumes that the clustered data points were drawn from an unknown union of multiple subspaces, have been proposed in recent years. Despite their successes in computer vision and data mining, most neglect to simultaneously consider global and local information, which may improve clustering performance. On the other hand, the number of connected components reflected by the learned affinity matrix is commonly inconsistent with the true number of clusters. To this end, we propose an adaptive affinity matrix learning method, nonnegative self-representation with a fixed rank constraint (NSFRC), in which the nonnegative self-representation and an adaptive distance regularization jointly uncover the intrinsic structure of data. In particular, a fixed rank constraint as a prior is imposed on the Laplacian matrix associated with the data representation coefficients to urge the true number of clusters to exactly equal the number of connected components in the learned affinity matrix. Also, we derive an efficient iterative algorithm based on an augmented Lagrangian multiplier to optimize NSFRC. Extensive experiments conducted on real-world benchmark datasets demonstrate the superior performance of the proposed method over some state-of-the-art approaches.

Keeping regional railway transportation safe is of great importance for railway system engineers and decision makers. However, there are still great challenges in modeling the complicated conditions in regional railway transportation: (1) Multiple types of data and knowledge in complicated correlations need to be analyzed, and (2) The approach must be open and accessible to decision makers so that a balanced decision can be made. To address the above challenges, a safety assessment approach using the hybrid Belief Rule Base (BRB) is proposed. In the new approach, multiple types of information are modeled under the hybrid assumption, and thus, hybrid rules are constructed to form the hybrid BRB. With this, both data and knowledge in complicated correlations can be used for the safety assessment on regional railway transportation, rather than only a single railway station or equipment component. Moreover, the assessment process remains open and accessible which provides good interpretability to stakeholders. An empirical regional railway safety assessment case is studied on the existing line and high speed line in the Cheng-Yu region located in the southwestern China. Five aspects, namely, the environment, equipment, management, passengers, and accident, are analyzed and then disintegrated into sub-factors. With the aspects and sub-factors, a comprehensive model is constructed. Case study results show that (1) the overall safety levels of the high speed line are better than the existing line, (2) the safety assessment results are consistent with the historical reports of accidents and system failures, (3) among all aspects, the environment and equipment have a more direct effect on the overall safety levels, and (4) consistency has also been found with railway accident statistics collected from Japan and Canada.

Recently, the deployment of sensors and actuators to collect and disseminate data in various applications such as e-healthcare, vehicular adhoc networks (VANETs) and smart factories has revolutionized several new communication technologies. The Internet of Things (IoT) is one of those emerging communication technologies. These revolutionary applications of IoT in industrial environment are termed as Industry 4.0 and it has vitalized the concept of Industrial IoT (IIoT). Being wireless communication, the authentication and integrity of data are the most important challenges. To mitigate these challenges, several digital signature schemes are proposed in the literature. However, due to identity-based or certificate-less construction, those schemes suffer from inborn key escrow and secret key distribution problems. To resolve such issues, the first certificate-based proxy signature (PFCBPS) scheme without pairing is proposed. The proposed PFCBPS scheme is provably secure in random oracle model (ROM). The performance comparison (in terms of computational costs of different phases and length of resulting delegation and signature) shows that the proposed PFCBPS scheme’s total computational cost is 46.69 msec. which is 52.24% of [8], 61.40% of [5], 23.33% of [20], 28% of [9] and 36.84% of [23]. Thus, it is more suitable to IIoT environment than existing competitive schemes.

In the article, we establish two multiple attribute decision making (MADM) approaches using the developed weighted generalized Maclaurin symmetric mean (q-ROFWGMSM) and weighted generalized geometric Maclaurin symmetric mean (q-ROFWGGMSM) operator concerning q-rung orthopair fuzzy numbers (q-ROFNs). Firstly, inspired by the generalized Maclaurin symmetric mean (G-MSM) and geometric Maclaurin symmetric mean (Geo-MSM) operators, we establish the q-rung orthopair fuzzy G-MSM (q-ROFGMSM) and q-rung orthopair fuzzy Geo-MSM (q-ROFGGMSM) operators, which assumes the grades of membership and non-membership to evaluate information can take any values in interval [0,1] respectively and the attributes are relevant to other multiple attributes. Then, we present its characteristics and some special cases. Moreover, we propose the weighted forms of the q-ROFGMSM and q-ROFGGMSM operator, which is called the q-ROFWGMSM and q-ROFWGGMSM operators, respectively. Then, we present their some characteristics and special examples. Finally, we put forward two new MADM approaches founded on the developed q-ROFWGMSM and q-ROFWGGMSM operators. The developed approaches are more general and more practicable than Liu and Wang's MADM approach (2018), Wei and Lu's MADM method (2017), Qin and Liu's MADM method (2014) and Shen et al.’s MADM approach (2018).

This paper studies the problem of adaptive fuzzy asymptotical quantized tracking control of non-strict-feedback systems with unmodeled dynamics. A dynamic signal is used to cope with the unmodeled dynamics and fuzzy systems are introduced to approximate the packaged unknown nonlinearities. Based on backstepping technique and fuzzy approximation property, a systemic fuzzy adaptive control scheme is proposed. By the utilization of Lyapunov theory, the semi-globally uniformly ultimate boundedness of all closed-loop system signals and asymptotical tracking performance are guaranteed. The main contributions of this work are two aspects: (i) a backstepping-based quantized control algorithm is firstly extended to nonlinear systems with unmodeled dynamics and non-strict-feedback structure; (ii) the semi-globally asymptotic tracking control scheme is independent of the quantized parameter. Simulation results verify the presented control approach.

As an important method of risk control in information systems and networks, cyber-insurance has attracted particular attention from both industry and academia. However, two prominent problems hamper the further growth of cyber-insurance. The correlated and interdependent properties of cyber-risks increase the economic risk of insurance companies considerably ; risk pooling can be impeded by these two properties. Further, this situation can be aggravated because cyber-insurance affects the investment for self-protection negatively. This phenomenon is regarded as the ex ante moral hazard. In this study, we establish a mathematical model based on a classic insurance theory to address the abovementioned problems, and propose an optimal cyber-insurance contract scheme that maximizes the expected utility of users. We also propose two personalized contract schemes to incentivize users to invest in self-protection under the no moral hazard and ex ante moral hazard conditions. Extensive experiments are conducted to evaluate the proposed approach, and the experimental results demonstrate the effectiveness and efficiency of the approach.

Mobile Crowdsensing is an emerging and promising sensing paradigm in which sensor data can be collected by mobile users equipped with smart devices. In Mobile Crowdsensing Systems (MCS), workers bid for location-based sensing tasks and get rewards from the platform. However, the bidding may leak workers’ path privacy, which means the sensitive locations could be inferred from innocent locations along a path as workers continuously acquire for tasks. This privacy concern may significantly hinder the participation of workers. As a result, this paper designs a novel framework for adjustable path privacy preservation used for task acquisition in MCS. In this framework, workers are allowed to flexibly adjust their privacy preferences on the amount, sensitivity, and cost of private locations. Two algorithms are proposed to determine the set of bidding tasks for workers that jointly consider the privacy concerns and profits. The first algorithm processes in a centralized approach, which is proved to be rational, truthful and efficient. The second algorithm allows workers to decide their task acquisition locally, and guarantees the Nash equilibrium among workers. Both algorithms are validated via real-world dataset. The evaluation results demonstrate that the two proposed algorithms outperform baseline algorithms on both platform and worker sides.

In this study, a new method of classification of skeleton-based motion data has been introduced. In the first stage, we performed multiscale feature extraction of rotational data. It is based on the proposed linear quaternion lifting scheme, with respect to the rotations coded by unit quaternions, which computes each scale based on the spherical linear interpolation (SLERP) prediction function and preserves the average signal value on each scale. Consequently, motion descriptors are extracted as quaternion attributes on different scales. The final recognition is performed by the nearest neighbor and minimum distance classifiers, adapted to support nonscalar features. Because of dimensionality of obtained descriptors, an attribute selection with respect to the multiresolution data has been proposed. It takes into consideration a specified number of resolutions, which is similar to low-pass filtering of the frequency domain. This method is utilized to solve the gait-based human identification problem. To validate such an application, a database containing data from 30 subjects was collected at the Human Motion Laboratory of the Polish-Japanese Academy of Information Technology (PJAIT). The obtained results were found to be satisfactory. In the best case, over 96% precision with only seven misclassified gaits of 178 samples was achieved.

Association rule mining (ARM) is largely employed in several scientific areas and application domains, and many different algorithms for learning association rules from databases have been introduced. Despite the presence of many existing algorithms, there is still room for the introduction of novel approaches tailored for novel kinds of datasets. Because often the efficiency of such algorithms depends on the type of analyzed dataset. For instance, classical ARM algorithms present some drawbacks for biological datasets produced by microarray technologies in particular containing Single Nucleotide Polymorphisms (SNPs). In particular classical algorithms require large execution times also with small datasets. Therefore the possibility to improve the performance of such algorithms by leveraging parallel computing is a growing research area. The main contributions of this paper are: a comparison among different sequential, parallels and distributed ARM techniques, and the presentation of a novel ARM algorithm, named Balanced Parallel Association Rule Extractor from SNPs (BPARES), that employs parallel computing and a novel balancing strategy to improve response time. BPARES improves performance without loosing in accuracy as well as it handles more efficiently the available computational power and reduces the memory consumption.

A recommender system utilizes information filtering techniques to help users obtain accurate information effectively and efficiently. The existing recommender systems, however, recommend items based on the overall ratings or the click-through rate, and emotions expressed by users are neglected. Conversely, the cold-start problem and low model scalability are the two main problems with recommender systems. The cold-start problem is encountered when the system lacks initial rating. Low model scalability indicates that a model is incapable of coping with high-dimensional data. These two problems may mislead the recommender system, and thus, users will not be satisfied with the recommended items. A hybrid recommender system is proposed to mitigate the negative effects caused by these problems. Additionally, ontologies are applied to integrate the extracted features into topics to reduce dimensionality. Topics mentioned in the items are displayed in the form of a topic map, and users can refer to these similar items for further information.

Sentiment classification is an important research task in Natural Language Processing. To fulfill this type of classification, previous works have focused on leveraging task-specific features. However, they only notice part of the related features. Also, state-of-the-art methods based on neural networks often ignore traditional features. This paper proposes a novel text sentiment classification method that learns the representation of texts by hierarchically incorporating multiple features. More specifically, we design different representations for sentiment words according to the polarity of labeled texts and whether negation exists; we distinguish words with different part-of-speech tags; emoticons, if there are, are to optimize the word vectors obtained in the previous step; apart from word embeddings, character embeddings are also trained. We use a deep neural network to get a sentence-level representation from both word and character sequence. For documents with at least two sentences, we use a hierarchical structure and design a rule to give more weight to import sentences empirically to get a document-level representation. Experimental results on open datasets demonstrate that our method could effectively improve the sentiment classification performance compared with the basic models and state-of-the-art methods.

Nowadays, cloud computing has developed well and been applied in many kinds of areas. However, privacy is still the most challenging problem which obstructs it being applied in some privacy-sensitive fields, such as finance and government. Advanced cryptographic algorithms provide data privacy with encryption, which can also support computation on such encrypted data. However, new challenge arises when such ciphertexts come from different parties. In particular, how to execute collaboratively data mining on encrypted data coming from different parties is a key issue from cloud service point of view. This paper focuses on privacy problem on outsourced k-means clustering scheme for two parties. In particular, each party’s data are encrypted only once and then stored in cloud. The proposed privacy-preserving k-means collaborative clustering protocol is executed mainly at the cloud, with O(k(m+n)) rounds of interactions among the two parties and the cloud, where m and n represent the total numbers of records for the two parties, respectively. It is shown that the protocol is secure in the semi-honest security model and in the malicious model in which only one party is corrupted during the process of centroids re-computation. Both theoretical and experimental analysis of the proposed scheme are also provided.

This paper investigates the Takagi-Sugeno (T-S) fuzzy based sliding-mode control (SMC) design of the discrete-time nonlinear model. By constructing a suitable fuzzy membership functions (FMFs) dependent Lyapunov function, the sufficient conditions are derived such that the resultant discrete-time T-S fuzzy model can achieve strictly (Q,S,R)−γ dissipative, where Q, S and R are known matrices with compatible dimensions satisfying Q≤0 and R=RT, and γ is a positive constant. Then, the desired control gain can be obtained by solving a set of linear matrix inequalities (LMIs). Besides that, a fuzzy SMC is designed to assure reaching condition. A modified fuzzy sliding-mode controller is also constructed to adapt input saturation. Finally, simulation results are presented to demonstrate the applicability and effectiveness of the proposed approaches.

This paper has unfolded from the study on rough sets, mainly from pre-rough algebra. Here we compare between implications of Łukasiewicz (Monteiro) 3-valued algebra and pre-rough algebra. This study aids in presenting an alternative axiomatization of Łukasiewicz (Monteiro) 3-valued algebra.

One of the most important topics of research in fuzzy sets is to get an appropriate notion of fuzzy metric space (FMS), in the paper we propose a new FMS–tripled fuzzy metric space (TFMS), which is a new generalization of George and Veeramani’s FMS. Then we present some related examples, topological properties, convergence of sequences, Cauchy sequence (CS) and completeness of the TFMS. Moreover, we introduce two kinds of notions of generalized fuzzy ψ-contractive (Fψ-C) mappings, and derive a fixed point theorem (FPT) on the mappings in the space.

Focusing on the problem of position tracking control for permanent magnet synchronous motors (PMSMs) with full-state constraints, this article proposes an adaptive fuzzy control scheme based on command filtering error compensation mechanism. Firstly, the unknown nonlinear functions of PMSM drive systems are approximated by utilizing the fuzzy logic systems (FLSs). Then, the command filtering technique is employed to deal with the “explosion of complexity” problem arising from conventional backstepping scheme, and the filtering errors are reduced by the error compensation mechanism. In addition, the barrier Lyapunov functions (BLFs) are constructed to guarantee that the state variables are restricted in compact bounding sets. Finally, simulation results show the effectiveness of the proposed scheme.

Many real-world networks own the characteristic of anti-community structure, i.e. disassortative structure, where nodes share no or few connections inside their groups but most of their connections outside. Detecting anti-community structure can explore negative relations among objects. However, the structures output by the existing algorithms are unbalanced, leading to no or few negative relations to be explored in some groups. Stochastic block models are promising methods for exploring disassortative structures in networks, but their results are highly dependent on the observed structure of a network. In this paper, we first improve the classic stochastic block model and propose a No sElf-edge Stochastic blOck Model (NESOM) for anti-community structure. NESOM considers the edges inside and among groups, respectively, and evolves a new objective function for evaluating anti-community structure. And then, a new heuristic algorithm NESOM-AC is proposed for balanced anti-community detection, which consists of three stages: creation of initial structure, decomposition of redundant group, and adjustment of group membership. Inspired by NESOM, we finally develop a new synthetic benchmark NESOM-Net for performance comparison. Experimental results on NESOM-Net with up to 100000 nodes and 16 real-world networks demonstrate the effectiveness of NESOM-AC in anti-community detection when compared with five state-of-the-art algorithms.

In order to obtain the weights of a set of criteria by means of real-world data, an effective method based on the covering-based variable precision intuitionistic fuzzy rough set (CVPIFRS) models is presented. By combining the CVPIFRS models with the idea of TOPSIS, we propose a decision-making method to effectively settle the complex and changeable bone transplant selections, which is one of typical multi-attribute decision-making (MADM) problems. The sensitivity analysis of the proposed method shows that the approach is highly flexible and can be applied to a wide range of environments by adjusting the values of the intuitionistic fuzzy (IF) variable precision, together with the choice of different IF logical operators. Through a comparison of the proposed method and some existing MADM methods, it is shown that our method is more effective in dealing with these complex and changeable bone transplant selections issues.

Revocable identity based signature (RIBS) is a useful cryptographic primitive, which provides a revocation mechanism to revoke misbehaving or malicious users over ID-based public key settings. In the past, many RIBS schemes have been previously proposed, but the security of all these existing schemes is based on traditional complexity assumptions, which are not secure against attacks in the quantum era. Lattice-based cryptography has many attractive features and it is all believed to be secure against attacks of quantum computing. Recently, Hung et al. proposed a RIBS with short size over lattices. However, in their scheme, it requires the private key generator (PKG) to perform linear work in the number of users and does not scale well. Moreover, their scheme is secure in the random oracle model. In this paper, we adopt the binary tree structure to present a scalable lattice-based RIBS scheme which greatly reduces the PKG’S workload associated with users from linear to logarithm. We prove that our proposed scheme is existentially unforgeable against chosen message attacks (EUF-CMA) under standard short integer solutions (SIS) assumption, in the standard model. Compared with the existing RIBS schemes over lattices, our proposed RIBS construction is secure in the standard model with scalability and meanwhile has efficient revocation mechanism with public channels.

With the proliferation of deep learning techniques, a significant number of applications related to home care systems have emerged recently. In particular, detecting abnormal events in a smart home environment has been extensively studied. In this paper, we adopt deep learning techniques, including convolutional neural networks (CNNs) and long short-term memory (LSTM) networks, to construct deep networks to learn the long-term dependencies from videos for human behavior recognition in a multiview framework. We adopt two cameras as our sensors to efficiently overcome the problem of occlusions and contour ambiguity for improving the accuracy performance of the multiview framework. After performing a series of image preprocessing on the raw data, we obtain human silhouette images as the input to our training model. In addition, because real-world datasets are complicated for analysis, labeling data is time consuming. Therefore, we present an image clustering method based on a stacked convolutional autoencoder (SCAE), which generates clustering labels for autolabeling. Finally, we set up our experimental environment as a normal residence to collect a large dataset, and the experimental results demonstrate the novelty of our proposed models.

In CloudIoT platform, the data is collected and shared by different nodes of Internet of Things (IoT), and data is processed and stored based on cloud servers. It has increased the abilities of IoT on information computation. Meanwhile, it also has enriched the resource in cloud and improved integration of the Internet and human world. All of this offer advantages as well as the new challenges of information security and privacy protection. As the energy limitation of the nodes in IoT, they are particularly vulnerable. It is much easier to hijack the nodes than to attack the data center for hackers. Thus, it is a crucial and urgent issue to realize the trusted update of authorization of nodes. When some nodes are hijacked, both of the behaviors to upload data to servers and to download information from servers should be forbidden. Otherwise, it might cause the serious damage to the sensitive data and privacy of servers. In order to solve this problem, we proposed a Proxy Re-encryption based Trusted Authorization scheme for nodes on CloudIoT (PRTA). PRTA is based on the proxy re-encryption (PRE), and the cloud server will play the roles of data storing and re-encrypting, which would reach the full potential of cloud computing and reduce the cost of nodes. The node’s status is taken as one of the parameters for data re-encryption and it is under the authorization servers’ control, which could ensure the security and reliability of the data and be beneficial for the privacy protection in CloudIoT. Also, the authorization servers are divided into the downloading and uploading kinds, which will make the application range much wider.

With the flourishing of ubiquitous healthcare and cloud computing technologies, medical primary diagnosis system, which forms a critical capability to link big data analysis technologies with medical knowledge, has shown great potential in improving the quality of healthcare services. However, it still faces many severe challenges on both users’ medical privacy and intellectual property of healthcare service providers, which deters the wide adoption of medical primary diagnosis system. In this paper, we propose an efficient and privacy-preserving medical primary diagnosis framework (CAMPS). Within CAMPS framework, the precise diagnosis models are outsourced to the cloud server in an encrypted manner, and users can access accurate medical primary diagnosis service timely without divulging their medical data. Specifically, based on partially decryption and secure comparison techniques, a special fast secure two-party vector dominance scheme over ciphertext is proposed, with which CAMPS achieves privacy preservation of user’s query and the diagnosis result, as well as the confidentiality of diagnosis models in the outsourced cloud server. Through extensive analysis, we show that CAMPS can ensure that users’ medical data and healthcare service provider’s diagnosis model are kept confidential, and has significantly reduce computation and communication overhead. In addition, performance evaluations via implementing CAMPS demonstrate its effectiveness in term of the real environment.

Secure computation outsourcing in Internet of Things (IoT) system is an ongoing research challenge, partly due to the resource-constrained nature of most (inexpensive) IoT devices. In this paper, we focus on the secure outsourcing of bilinear pairings (SOBP) (the most computationally expensive operation in pairing-based cryptographic protocols / algorithms). First, we analyze the limitations in existing SOBP-based schemes, such as the one-malicious model (Strong Assumption), a secure channel (Insufficiency), and a trusted server (Centralization). Then, we propose a novel blockchain-based system for SOBP based on a permissioned version (i.e., a blockchain ledger maintained by some permissioned nodes), designed to efficiently address the limitations. Finally, we prove the security of our proposed approach in the one untrusted program model and implement it on Ethereum (an open-source blockchain system) to show its utility.

The theory of Markowitz portfolio has had enormous value and extensive applications in finance since it came into being. With the advent of the Big-Data era and the increasingly complicated financial market, the resource consumption of computing portfolio investments is significantly increasing. Cloud computing offers a good platform to efficiently compute large-scale portfolio investments, in particular, for resource-limited investors. In this paper, a Markowitz model (MM) is taken into consideration for outsourcing to a public cloud in a privacy-conscious way. As in general computation outsourcing, outsourcing MM inevitably faces four issues, namely, input/output privacy, correctness, verification, and substantial computation gain for investors; it has consistent complexity with the original methods when the cloud solves the encrypted version. However, the proposed cloud-assisted privacy-conscious MM employs location-scrambling and value-alteration encryption operations, which can protect the MM’s input/output privacy well. Moreover, the correctness of solving MM over an encrypted domain in the cloud side can be demonstrated and the results returned by the cloud can be verified. Furthermore, both theoretical and experimental analyses validate that the investor can obtain a huge amount of computational gain, and the cloud complexity consistent with that of the original case when solving the encrypted version.

Privacy preservation has become a prerequisite for modern applications in the cloud, social media, Internet of things (IoT), and E- healthcare systems. In general, health and medical data contain images and medical information about the patients and such personal data should be kept confidential in order to maintain the patients’ privacy. Due to limitations in digital data properties, traditional encryption schemes over textual and structural one-dimension data cannot be applied directly to e-health data. In addition, when personal data are sent over the open channels, patients may lose privacy of data contents. Hence, a secure lightweight keyframe extraction method is highly required to ensure timely, correct, and privacy-preserving e-health services. Besides this, it is inherently difficult to achieve a satisfied level of security in a cost-effective way while considering the constraints of real-time e-health applications. In this paper, we propose a privacy preserving chaos-based encryption cryptosystem for patients’ privacy protection. The proposed cryptosystem can protect patient’s images from a compromised broker. In particular, we propose a fast probabilistic cryptosystem to secure medical keyframes that are extracted from wireless capsule endoscopy procedure using a prioritization method. The encrypted images produced by our cryptosystem exhibits randomness behavior, which guarantee computational efficiency as well as a highest level of security for the keyframes against various attacks. Furthermore, it processes the medical data without leaking any information, thus preserving patient’s privacy by allowing only authorized users for decryption. The experimental results and security analysis from different perspectives verify the excellent performance of our encryption cryptosystem compared to other recent encryption schemes.

How to predict the future behavior of complex systems with insufficient information, i.e., low dimensional mid-term chaotic time series prediction in mathematical terms, is not only a significant theoretical problem, but a more intricate practical problem. To address this issue, a Delay Parameterized Method (DPM) for low dimensional mid-term chaotic time series forecasting is presented. The correlation function, which immerses the low dimensional information into reconstructed space, is introduced to bridge time series and hidden order of system in DPM. Traversal algorithm and intelligent algorithm including particle swarm optimization or genetic algorithm, are used to obtain the optimal parameters for prediction. In addition, the applications of the proposed method on Lorenz chaotic time series, stress-strain signals and stock K-line maps show that it produces high quality predictions.

Data stream mining has gained increasing attention in recent years due to its wide range of applications. In this paper, we propose a new selective prototype-based learning (SPL) method on evolving data streams, which dynamically maintains representative instances to capture the time-changing concepts, and make predictions in a local fashion. As an instance-based learning model, SPL only maintains some important prototypes (i.e., ISet) via error-driven representativeness learning. The fast condensed nearest neighbor (FCNN) rule, is further introduced to compress these prototypes, making the algorithm also applicable under memory constraints. To better distinguish noises from the instances associated with the new emerging concept, a potential concept instance set (i.e., PSet) is used to store all misclassified instances. Relying on the potential concept instance set, a local-aware distribution-based concept drift detection approach is proposed. SPL has several attractive benefits: (a) it can fit the evolving data streams very well by maintaining a small size of instance set; (b) it is capable of capturing both gradual and sudden concept drifts effectively; (c) it has great capabilities to distinguish noise/outliers from drifting instances. Experimental results show that the SPL has better classification performance than many other state-of-the-art algorithms.

Edge cloud computing can provide resources that are close to users and reduce response time. However, the edge cloud computing system still faces many challenges on addressing the overload problem due to its limited capacity. In this paper, a resource management strategy is proposed to satisfy the workloads of the edge cloud, while minimizing the financial cost of the rented nodes. Furthermore, the replica management strategy, which consists of the replica allocation and consistency preservation strategy, is studied. A dynamic replica allocation strategy is proposed to satisfy the user experience while reducing the storage overheads. In addition, the replica consistency preservation strategy is proposed for guaranteeing the data consistency and correctness. Finally, extensive experiments are conducted based on a real-world dataset. With the increase in the time, the proposed resource management algorithm can significantly reduce the total financial cost of the rented nodes and SLA default rate and improve the CPU utilization. For instance, the total financial cost of the proposed algorithm averagely achieves up to 32.27% and 53.65% reduction over that of CAAS algorithm and DRM algorithm, respectively. In addition, the proposed replica allocation algorithm can effectively reduce the data transmission time and the storage overhead.

This study proposes an optimization framework based on a hybrid information-entropy approach to identify leakage events in water distribution networks (WDN). Optimization-based methods are widely employed in the literature for such purposes; however, they are constrained by time-consuming procedures. Hence, researchers eliminate parts of the decision space to curtail the computational burden. Here, we propose an information theory-based approach, using Value of Information (VOI) and Transinformation Entropy (TE) methods, in conjunction with an optimization model to explore the entire decision space. VOI allows for the entire feasible space search through intelligent sampling, which in turn ensures robust solutions. TE minimizes redundant information and helps maximize the spatial distribution of sensors. The herein proposed model is developed within a multi-objective optimization framework that renders a set of Pareto-optimal solutions. ELimination and Choice Expressing the REality (ELECTRE) multi-criteria decision-making model is then used to select the best compromise solution given several weighting scenarios. The results of this study show that the information-entropy based scheme can improve the precision of leak detection by enhancing the decision space, and can reduce the computational burden.

In the age of big data explosion, the management of online public opinion has encountered great challenges. Which way can effectively manage online public opinion has become a decision-making question for us to think about. Probabilistic uncertain multiplicative linguistic preference relations (PUMLPRs) are a remarkable instrument to solve uncertain evaluation problems. This paper uses the PUMLPRs to assess the management ways of the online public opinion. Owing to the intricacy of decision-making domain, the PUMLPRs are not always complete. We get the complete PUMLPRs in two steps: the repair for uncertain multiplicative linguistic variables and the repair for probability. Moreover, the consistency of the complete PUMLPRs is researched. Then the final priorities are obtained by the proposed possibility degree formula. After that, the numerical example that helps assess the valid way to manage online public opinion is performed to check the feasibility of the proposed decision-making procedure.

Public key encryption with equality test (PKEET) is a cryptosystem that allows a tester who has trapdoors issued by one or more users Ui to perform equality tests on ciphertexts encrypted using public key(s) of Ui. Since this feature has a lot of practical applications including search on encrypted data, several PKEET schemes have been proposed so far. However, to the best of our knowledge, all the existing proposals are proven secure only under the hardness of number-theoretic problems and/or the random oracle heuristics. In this paper, we show that this primitive can be achieved not only generically from well-established other primitives but also even without relying on the random oracle heuristics. More precisely, our generic construction for PKEET employs a two-level hierarchical identity-based encryption scheme, which is selectively secure against chosen plaintext attacks, a strongly unforgeable one-time signature scheme and a cryptographic hash function. Our generic approach toward PKEET has several advantages over all the previous works; it directly leads the first standard model construction and also directly implies the first lattice-based construction. Finally, we show how to extend our approach to the identity-based setting.

Compared with the traditional classification task based on a single image, an image set contains more complementary information, which is of great benefit to correctly classify a query subject. Thus, image set classification has attracted much attention from researchers. However, the main challenge is how to effectively represent an image set to fully exploit the latent discriminative feature. Unlike in previous works where an image set was represented by a single or a hybrid mode, in this paper, we propose a novel multi-model fusion method across the Euclidean space to the Riemannian manifold to jointly accomplish dimensionality reduction and metric learning. To achieve the goal of our framework, we first introduce three distance metric learning models, namely, Euclidean-Euclidean, Riemannian-Riemannian and Euclidean-Riemannian to better exploit the complementary information of an image set. Then, we aim to simultaneously learn two mappings performing dimensionality reduction and a metric matrix by integrating the two heterogeneous spaces (i.e., the Euclidean space and the Riemannian manifold space) into the common induced Mahalanobis space in which the within-class data sets are close and the between-class data sets are separated. This strategy can effectively handle the severe drawback of not considering the distance metric learning when performing dimensionality reduction in the existing set based methods. Furthermore, to learn a complete Mahalanobis metric, we adopt the L2,1 regularized metric matrix for optimal feature selection and classification. The results of extensive experiments on face recognition, object classification, gesture recognition and handwritten classification demonstrated well the effectiveness of the proposed method compared with other image set based algorithms.

With the rapid development of information technology, information is generated at an unprecedented rate. Users are in great need of recommender systems to provide the potential friends or interested items for them. Social (i.e. friend) recommendation and behavior (i.e. item) recommendation are two types of popular services in real-world applications. Although researchers have proposed various models for each task, a unified model to address both tasks elegantly and effectively is still in demand. In this paper, we propose a model called SBRNE which integrates social and behavior recommendations into a unified framework through modeling social and behavior information simultaneously. Specifically, SBRNE models social and behavior information simultaneously via employing users’ latent interests as a bridge, and derives improved performance on both social and behavior recommendation tasks. In addition, by introducing an efficient network embedding procedure, users’ latent representations are advanced, and effectiveness and efficiency of recommendation tasks are improved accordingly. Results on both real-world and synthetic datasets demonstrate that: 1). SBRNE outperforms selected baselines on social and behavior recommendation tasks; 2). SBRNE performs stable on recommendation tasks for cold-start users; 3). The network embedding procedure can improve the effectiveness of SBRNE; 4). The hyper-parameter learning procedure can improve both the effectiveness and efficiency of SBRNE.