Search

In this survey, a systematic literature review of the state-of-the-art on emotion expression recognition from facial images is presented. The paper has as main objective arise the most commonly used strategies employed to interpret and recognize facial emotion expressions, published over the past few years. For this purpose, a total of 51 papers were analyzed over the literature totaling 94 distinct methods, collected from well-established scientific databases (ACM Digital Library, IEEE Xplore, Science Direct and Scopus), whose works were categorized according to its main construction concept. From the analyzed works, it was possible to categorize them into two main trends: classical and those approaches specifically designed by the use of neural networks . The obtained statistical analysis demonstrated a marginally better recognition precision for the classical approaches when faced to neural networks counterpart, but with a reduced capacity of generalization . Additionally, the present study verified the most popular datasets for facial expression and emotion recognition showing the pros and cons each and, thereby, demonstrating a real demand for reliable data-sources regarding artificial and natural experimental environments.

Traditional topic classification usually adopts the closed-world assumption that all the test topics have been seen in training. However, in open dynamic environments, the potential new topics may appear in testing due to the evolution of text data over time. Considering the uncertainty and multi-granularity of dynamic text data, such open topic classification needs to detect unseen topics by mining the boundary region continually, and incrementally update the previous models by knowledge accumulation. To address these challenge issues, this paper introduces a unified framework of three-way multi-granularity learning to open topic classification based on the fusion of three-way decision and granular computing. First, we propose the multilevel granular structure of tasks from the temporal-spatial multi-granularity perspective. Then, we construct an adaptive decision boundary and use the centroids and the corresponding radius to discover unknowns by the reject option. Subsequently, we further explore the unknown topics by three-way enhanced clustering and the uncertain instances will be re-investigated in the next stage. Besides, we design a built-in knowledge base represented as the centroid of each topic to store the topic knowledge. Finally, the experiments are conducted to compare the performances of proposed models and the efficiency of knowledge accumulation with classic models.

Aspect-level sentiment classification aims to obtain fine-grained sentiment polarities of different aspects in one sentence. Most existing approaches handle the classification by acquiring the importance of context words towards each given aspect individually, and ignore the benefits brought by aspect relations. Since the sentiment of one aspect can be deduced through their relationship according to other aspects, in this paper, we propose a novel relation construction multi-task learning network (RMN), which is the first attempt to extract aspect relations as an auxiliary classification task. RMN generates aspect representations through graph convolution networks with a semantic dependency graph and utilizes the bi-attention mechanism to capture the relevance between the aspect and the context. Unlike conventional multi-task learning methods that need extra datasets, we construct an auxiliary relation-level classification task that extracts aspect relations from the original dataset with shared parameters. Extensive experiments on five public datasets from SemEval 14, 15, 16 and MAMS show that our RMN improves about 0.09% to 0.8% on accuracy and about 0.04% to 1.19% on F1 score, compared to several comparative baselines.

In this paper, the issues of both fault estimation and accommodation are studied for a class of continuous-time Markov jump linear systems under actuator fault, sensor fault and external disturbance in the framework of finite frequency domain. The imprecise statistic of modes transitions are considered here, which means that the transition rates are uncertain. Firstly, the sensor fault is defined as a new state, and then an adaptive observer is developed to estimate the actuator fault and the newly defined state with the aid of the descriptor system approach. By using the obtained fault estimation, a novel fault accommodation scheme is proposed based on the sample point controller design approach to satisfy the stochastic stability requirement of the closed-loop system with a prescribed H ∞ performance bound. Moreover, for the convenience of design, the presented sample point controller design is successfully transformed into resolving a class of input-delay control issues. Finally, two numerical examples, including a practical F-404 aircraft engine system, are illustrated to show the effectiveness and applicability of the developed design methods.

Time-series data are widespread in real-world industrial scenarios. To recover and infer missing information in real-world applications, the problem of time-series prediction has been widely studied as a classical research topic in data mining. Deep learning architectures have been viewed as next-generation time-series prediction models. However, recent studies have shown that deep learning models are vulnerable to adversarial attacks. In this study, we prospectively examine the problem of time-series prediction adversarial attacks and propose an attack strategy for generating an adversarial time series by adding malicious perturbations to the original time series to deteriorate the performance of time-series prediction models. Specifically, a perturbation-based adversarial example generation algorithm is proposed using the gradient information of the prediction model. In practice, unlike the imperceptibility to humans in the field of image processing, time-series data are more sensitive to abnormal perturbations and there are more stringent requirements regarding the amount of perturbations. To address this challenge, we craft an adversarial time series based on the importance measurement to slightly perturb the original data. Based on comprehensive experiments conducted on real-world time-series datasets, we verify that the proposed adversarial attack methods not only effectively fool the target time-series prediction model LSTNet, they also attack state-of-the-art CNN-, RNN-, and MHANET-based models. Meanwhile, the results show that the proposed methods achieve a good transferability. That is, the adversarial examples generated for a specific prediction model can significantly affect the performance of the other methods. Moreover, through a comparison with existing adversarial attack approaches, we can see that much smaller perturbations are sufficient for the proposed importance-measurement based adversarial attack method. The methods described in this paper are significant in understanding the impact of adversarial attacks on a time-series prediction and promoting the robustness of such prediction technologies.

Biometrics has nowadays become a preferred solution for systems requiring secure authentication. However, the usage of biometric characteristics raises significant concerns regarding personal data protection and privacy. Several template protection schemes have been therefore proposed to conceal the employed identifiers, while still ensuring the ability to efficiently recognise users. In this paper, we propose and analyse three different approaches generating cancelable templates from finger vein features. A thorough analysis of the considered methods is conducted to investigate their impact on the achievable recognition performance, as well as their security in terms of renewability and unlinkability. Furthermore, a specific attack is designed to evaluate the irreversibility of the protection scheme providing the best recognition performance.

The missing data problem - attributed to malfunctioning detectors, packet loss during transmission, or data removed by quality control procedures - is unavoidable in most traffic-related datasets. However, this problem has adversely affected traffic engineering applications as they heavily rely on accurate and comprehensive data. This study aims to impute missing loop detector data in order to improve the estimation results of traffic flow analysis. This paper presents a statistically principled methodology that focuses not only on proposing a computationally efficient imputation approach, but also on assessing the uncertainty associated with imputed values. The proposed methodology quantifies the accuracy of imputation and estimation of uncertainty for a range of challenging patterns of missing loop detector data, and compares them with existing methods. The results of the analysis demonstrate that the performance of the proposed approach remains unaffected by the presence of a large number of missing patterns and reflects the true statistical properties of the principal data. The proposed approach is also comparatively less computationally complex than the existing methods. Further, the comparative analysis of the proposed estimator shows that the generated prediction intervals are reasonably accurate and conform to the desired confidence levels with relatively small interval width.

Cloud service selection (CSS) remains a strategically significant decision and has a substantial impact on an organization’s competitive edge. Despite considerable research, the literature lacks a comprehensive unified approach to consensual CSS. Recognizing the significance of CSS decisions, in this paper, we propose a novel Cloud Service Scrutinization and Selection Framework (C3SF) that includes four interdependent phases: (1) requirements elicitation, (2) scrutinization, (3) evaluation, and (4) ranking & selection. As part of C3SF, we use conjunctive screening to scrutinize cloud services. We also propose a novel multi-criteria decision-making (MCDM) approach called the modified Best-Worst Method (MBWM), which computes the weights of criteria using early-stage consensus among decision-makers. In addition, we introduce an innovative two-step consensus process for ranking services using leading MCDM methods followed by an aggregation of ranks using a Markov chain-based approach. Moreover, to develop a broader consensus, we propose another two-stage novel mechanism comprising multi-aggregation and synthesis/fusion of rank information using a partially ordered set. We validate the performance and effectiveness of C3SF through a CSS case study using real-world data followed by a comprehensive analysis. The results show that C3SF is robust, practical, and suitable for well-informed decision-making.

The forecasting of time series provides great convenience in our daily life. Studies of time series forecasting have been used in many fields such as financial models, weather, and traffic patterns. In this paper, we propose a model fusion-based time series forecasting to improve the forecasting accuracy and efficiency. We propose a time series forecasting scheme based on a multivariate grey model and uses artificial fish swarm algorithm to optimize the settings. We then propose two fusion models with the grey model-based schemes on two different perspectives: data decomposition, and weighted summation. We conduct evaluations based on real data series and compared them with other forecasting models. Results show that our model can achieve good prediction accuracy and efficiency, which can be used for time series forecasting in different scenarios.

In recent years, many applications are using various forms of deep learning models. Such methods are usually based on traditional learning paradigms requiring the consistency of properties among the feature spaces of the training and test data and also the availability of large amounts of training data, e.g., for performing supervised learning tasks. However, many real-world data do not adhere to such assumptions. In such situations transfer learning can provide feasible solutions, e.g., by simultaneously learning from data-rich source data and data-sparse target data to transfer information for learning a target task. In this paper, we survey deep transfer learning models with a focus on applications to text data. First, we review the terminology used in the literature and introduce a new nomenclature allowing the unequivocal description of a transfer learning model. Second, we introduce a visual taxonomy of deep learning approaches that provides a systematic structure to the many diverse models introduced until now. Furthermore, we provide comprehensive information about text data that have been used for studying such models because only by the application of methods to data, performance measures can be estimated and models assessed. publishedVersion

Assisting end users to identify desired results from a large dataset is an important problem for multi-criteria decision making. To address this problem, top-k and skyline queries have been widely adopted, but they both have inherent drawbacks, i.e., the user either has to provide a specific utility function or faces many results. The k-regret minimization query is proposed, which integrates the merits of top-k and skyline queries. Due to the NP-hardness of the problem, the k-regret minimization query is time consuming and the greedy framework is widely adopted. However, formal theoretical analysis of the greedy approaches for the quality of the returned results is still lacking. In this paper, we first fill this gap by conducting a nontrivial theoretical analysis of the approximation ratio of the returned results. To speed up query processing, a sampling-based method, S toc P res G reed , is developed to reduce the evaluation cost. In addition, a theoretical analysis of the required sample size is conducted to bound the quality of the returned results. Finally, comprehensive experiments are conducted on both real and synthetic datasets to demonstrate the efficiency and effectiveness of the proposed methods.

This paper addresses the event-triggered control problem of discrete-time uncertain networked control systems under random deception attacks. In order to deal with a network resource constraint, a novel two-event-generator scheme over both sensor-to-controller and controller-to-actuator channels is proposed to reduce the transmission frequency of system states and control commands, while maintaining desired system performance. More specifically, one of the event-generators is a time-varying memory-based event-triggered scheme (METS), and the other is the traditional event-triggered scheme (ETS). On the other hand, in order to incorporate network security , a new model of memory-based NCSs is established under random deception attacks. Then, sufficient conditions for preserving the asymptotic stability of uncertain NCSs are proposed by using the Lyapunov functional method. Furthermore, co-design criteria are derived for determining both the memory-based state feedback controller gains and event-triggering parameter matrices. Finally, the effectiveness of the proposed methods is verified by a quarter-car suspension system.

This paper aims to solve sample inefficiency in Asynchronous Advantage Actor-Critic (A3C). First, we design a new off-policy actor-critic algorithm, which combines actor-critic with experience replay to improve sample efficiency. Next, we study the sampling method of experience replay for trajectory experience and propose a familiarity-based replay mechanism which uses the number of replay times of experience as the probability weight of sampling. Finally, we use the GAE-V method to correct the bias caused by off-policy learning. We also achieve better performance by adopting a mechanism that combines off-policy learning and on-policy learning to update the network. Our results on Atari and MuJoCo benchmarks show that each of these innovations contributes to improvements in both data efficiency and final performance. Furthermore, our approach keeps a fast coverage speed and the same parallel feature as A3C, and also has better performance on exploration.

Existing block-based progressive visual cryptography schemes (BPVCS) that prevent cheating are limited to ( 2 , n ) and ( k , n ) thresholds. They are unsuitable for implementing complex sharing techniques in the actual world. A cheating immune BPVCS for general access structure (GAS) is proposed in this paper. To find the parameters that make up the proposed scheme, a simplified multiple assignment procedure based on integer linear programming is introduced. The secret parts can be decrypted by the members of a qualified set. Participants in a forbidden set, on the other hand, are not allowed to know about the secret. The proposed scheme includes cheat-prevention and progressive recovery capabilities. In addition, to generate meaningful shadows for BPVCS, a modified extended VCS (EVCS) is proposed. Our modified EVCS can solve the cover-interference problem in conventional BPVCS. The effectiveness and benefits of the proposed BPVCS are supported through the experimental findings and comparisons.

Recommender system has recently received a lot of attention in the information service community. In many application scenarios, such as Internet of Things (IoTs) environments, item multimodal auxiliary information (such as text and image) can be obtained to expand their feature representation and to increase user satisfaction with recommendations. Motivated by this fact, this paper introduces a novel two-stage embedding model (TSEM), which adequately leverage item multimodal auxiliary information to substantially improve recommendation performance. Specifically, it encompasses two sequential stages: graph convolutional embedding (GCE) and multimodal joint fuzzy embedding (MJFE). In the former, we first generate a bipartite graph for user-item interactions, and then utilize it to construct user and item backbone features via a spatial-based graph convolutional network (SGCN). While in the latter, by employing item multimodal auxiliary information, we integrate multi-task deep learning, deterministic Softmax, and fuzzy Softmax into a convolutional neural network (CNN)-based learning framework, which is optimized to obtain user backbone features and item semantic-enhanced fuzzy (SEF) features accurately. After TSEM converges, user backbone features and item SEF features can be utilized to calculate user preferences on items via Euclidean distance . Extensive experiments over two real-world datasets show that the proposed TSEM model significantly outperforms the state-of-the-art baselines in terms of various evaluation metrics .

Recent works on salient object detection (SOD) mainly focus on pixel-level classification by leveraging fully convolutional network (FCN)-based encoder-decoder models. In this paper, considering that the context relation plays a critical role in defining a salient object appearing in a scene, we propose a local-to-global context-aware feature augmentation network, namely LGCNet. A two-branch attention-based context relation modeling structure is designed by considering global context-aware information based on foreground/background cues and global feature representations. A pixel-wise self-attention mechanism is then incorporated for both branches to propagate global context information to local feature representations. As a result, a coarse-to-fine salient object detection model is formulated. The whole framework can be trained in an end-to-end manner under a deeply supervised framework. Experimental results demonstrate the effectiveness of key components proposed in our LGCNet. Our LGCNet achieves promising results in comparison with 18 state-of-the-art methods on six widely-used benchmark datasets.

Multi-modal brain imaging data reflect brain structural and functional information from different aspects, which have been widely used in brain disease diagnosis, including epilepsy and Alzheimer's disease . In practice, it is difficult to obtain all the modalities of each subject due to high cost or equipment limitation. Therefore, it is highly essential to fuse incomplete multi-modality data to improve the diagnostic accuracy . The traditional methods need to perform data cleansing and discard incomplete subjects from the data, which leads to inefficient training and poor robustness. For addressing this problem, this paper proposes an incomplete multi-modality data fusion method based on low-rank representation for the diagnosis of epilepsy and its subtypes. Specifically, we designed an objective function that simultaneously learns the low-rank representation of the complete modality part, and recovers the incomplete modality by the correlation between different modalities. The proposed model can be optimized by using alternating direction method of multipliers . Extensive evaluation of the proposed method on epilepsy classification task with incomplete DTI and fMRI data showed that our method can achieve promising classification results in identifying epilepsy and its subtypes.

This paper investigates an event triggered control problem of connected vehicles under multiple cyber-attacks, including denial-of-service attacks (DoS) and deception attacks, which occur aperiodically. Then, a novel event triggered control scheme is proposed in the presence of multiple cyber-attacks. The upper bound of attack frequency, attack length rate of DoS attacks and the conditions of controller gains are obtained. It is proven that the proposed distributed control scheme can realize the asymptotic tracking performance of connected vehicles while avoiding Zeno behavior. A string stability analysis under multiple cyber-attacks is also provided. Simulation results confirm the effectiveness of the event triggered control schemes under multiple cyber-attacks.

The data collected from the real world are diverse and include categorical data , numerical data, incomplete data and noisy data. In addition, many real data sets may dynamically vary, and dynamic data display characteristics with multi-dimensional variations. However, for mixed incomplete data systems, most of the existing incremental methods only work well with single-dimensional dynamic data sets and are not suitable for processing specific multi-dimensional variations of objects and attributes. In this paper, we focus on researching dynamic approaches to efficiently update three-way regions based on the simultaneous variations of the object set and the attribute set in an incomplete neighborhood decision system (INDS). First, considering the complexity of data, we utilize matrix approaches to calculate three-way regions of the INDS based on a proposed neighborhood tolerance relation. Then, under the simultaneous addition of the object set and the attribute set in the INDS, we research incremental mechanisms based on the matrix to obtain three-way regions from previous knowledge. Subsequently, an incremental algorithm for updating three-way regions is proposed when the object set and the attribute set are simultaneously added to the INDS. Finally, the results of a series of experiments and comparisons based on UCI data sets show that the performance of the proposed incremental algorithm is much better than that of the traditional static algorithm, the integrated single-dimensional incremental algorithm and the single-level combined incremental algorithm.

In this study, we focus on semi-supervised data stream classification tasks. With the advent of applications that generate vast streams of data, data stream mining algorithms are required to handle the intrinsic issues of learning patterns from this type of non-stationary data. The main challenges are fast arriving, infinite length, limited labeled data, concept drift, and concept evolution. Although many algorithms have been used to overcome these issues, there are still many challenges that need to be addressed and discussed in practical domains. This paper mainly focuses on the limited labeled data and concept drift detection issues. Therefore, a semi-supervised framework called CPSSDS is proposed, which uses an incremental classifier as a base learner and a self-training framework to handle the scarcity of labeled samples. The proposed approach utilizes a form of Conformal Predictors to discover a set of informative unlabeled data instances to add to the original training set in each training procedure which is the main challenge in the standard self-training approach. In terms of this framework, the Kolmogorov-Smirnov test is adopted to detect concept drift by comparing the outputs of conformal prediction for two sequences of data chunks. Finally, experimental results on several synthetic and real datasets demonstrate that the proposed learning framework is comparable to the state-of-the-art semi-supervised algorithms for data streams. The implementation source code of the CPSSDS is available at https://github.com/neginsmd/CPSSDS.git .

Given that the three-way decision (3WD) theory provides a scientific solution and reasonable semantic interpretation for solving multi-criteria decision-making (MCDM) problems, this paper presents a new 3WD model and applies it to realistic MCDM problems. Since the fuzzy neighborhood operator is effective in handling uncertain data, we first define a pair of probabilistic approximation operators via the constructed α -fuzzy neigh borhood class. Then, a 3WD model is explored in decision information systems to address the related issues in the proposed fuzzy probabilistic rough set model. In light of the above works, we establish a novel approach to classify and rank applicants for enterprise talent recruitment problems. Instead of the rule of tie-breaking, a way is given to separate applicants into three parts by combining decision attributes , and the feasibility of the approach is confirmed as well. Finally, by using a data set in the UCI database, the results of comparative and experimental analyses demonstrate that the constructed MCDM approach owns better performances in terms of effectiveness and stability.

This paper investigates the fault-tolerant load frequency control (LFC) design of interconnected wind power systems . For this, a doubly-fed induction generator (DFIG)-based wind farm is integrated into the interconnected power systems . Distinct from the existing works, the communication time-delay is taken into the area control error (ACE) signals of interconnected wind power systems. Also, compared with the traditional LFC scheme, a fault-tolerant H ∞ LFC scheme is presented with known and unknown actuator fault against load disturbances and wind speed fluctuations. Then, the asymptotic stability conditions are derived in the linear matrix inequalities (LMIs) form by utilizing the Lyapunov–Krasovskii functional (LKF) method and Wirtinger-based inequality technique under the designed LFC scheme with H ∞ performance. Finally, numerical examples are given to validate the applicability and superiority of the designed LFC scheme.

Task scheduling is an important research direction in cloud computing. The current research on task scheduling considers mainly the design of scheduling strategies and algorithms and rarely gives attention to the influences of uncertain factors, such as the network bandwidth and millions of instructions per second (MIPS), on the scheduling process. The network bandwidth and MIPS directly affect the performance of a virtual machine (VM), which further influences the scheduling performance. In this paper, uncertain factors are transformed into interval parameters. The make-span, scheduling cost, load balance, and task completion rate are simultaneously considered in the scheduling process. Then, an interval many-objective cloud task scheduling optimization (I-MCTSO) model is designed to simulate real cloud computing task scheduling. To implement this model, an interval many-objective evolutionary algorithm (InMaOEA) is proposed. An interval credibility strategy is employed to improve the convergence performance. The hyper-volume and degree of overlap based on the interval congestion distance strategy are used to increase the population diversity. Simulation results demonstrate the effectiveness and superior performance of InMaOEA in comparision with other algorithms. The proposed approaches can provide decision-makers with an efficient allocation plan for cloud task scheduling.

As a powerful tool for data streams processing, the vast majority of existing evolving intelligent systems (EISs) learn prediction models from data in a supervised manner. However, high-quality labelled data can be difficult to obtain in many real-world classification applications concerning data streams, though unlabelled data is plentiful. To overcome the labelling bottleneck and construct a stronger classification model, a novel semi-supervised EIS is proposed in this paper. After being primed with a small amount of labelled data, the proposed method is capable of continuously self-developing its system structure and self-updating the meta-parameters from unlabelled data streams chunk-by-chunk in a non-iterative, exploratory manner by exploiting a novel pseudo-labelling strategy. Thanks to its transparent prototype-based structure and human-understandable reasoning process, the proposed method can provide users high explainability and interpretability while achieving great classification precision. Experimental investigation demonstrates the superior performance of the proposed method.

Many embedding approaches of drugs have been proposed for the downstream task of drug-drug interaction (DDI) prediction in a DDI-derived network where drugs are considered nodes, and interactions are represented as edges. One of the most popular approaches is learning the representation of a drug from the DDI network by aggregating the features or information of its neighboring drugs. However, existing methods do not consider the specific type of the relation between the drugs, leading to an incomplete embedding learning process. Given that different relations between drugs may have different effects on drug embedding, the combination of multirelational embedding and relation-aware network structure embedding of drugs can be helpful to improve the prediction of DDIs. Therefore, in this paper, a relation-aware network embedding model for the prediction of drug-drug interactions (RANEDDI) is proposed. RANEDDI not only considers the multirelational information between drugs but also integrates the relation-aware network structure information in the topology of a multirelational DDI network to obtain the drug embedding. Under evaluation metrics such as AUC, AUPR and F1, the experimental results show that RANEDDI is superior to several state-of-the-art methods and can be used in the prediction of binary and multirelational DDIs. We also perform ablation studies that demonstrate that RANEDDI is effective and that it is robust in the task of binary DDI prediction, even in the case of a scarcity of labeled DDIs. The source code is freely available at https://github.com/DongWenMin/RANEDDI.

Nowadays, an increasing amount of data has been sent to the cloud for analysis and storage, and data security in the cloud has been widely concerned. Among them, CP-ABE is regarded as one of the most promising technologies to protect outsourced data. However, in most CP-ABE schemes, attackers may obtain user privacy information from policy plaintext. More likely, partial policy hiding programme neither satisfies full policy hiding nor applies to the needs of the large universe. In this paper,we hide both attribute values and names in policy by using private set intersection (PSI). The CP-ABE programme not only supports a complete hiding policy, but also can calculate the authorization relationship and mapping relationship between user passwords and keys. We use a polynomial-based PSI and a recursive algorithm to calculate the authorization relationship. And with the help of the algorithm and the label vector, the mapping is determined under communication restrictions. Through the outsourcing index, we achieve an efficient hiding strategy and effectively reduce the user's computing overhead. Finally, theoretical analysis and experiments show that our model has better performance while effectively protecting sensitive attributes.

In the Internet age, analyzing the behavior of online users can help webstore owners understand customers’ interests. Insights from such analysis can be used to improve both user experience and website design. A prominent task for online behavior analysis is clickstream mining, which consists of identifying customer browsing patterns that reveal how users interact with websites. Recently, this task was extended to consider weights to find more impactful patterns. However, most algorithms for mining weighted clickstream patterns are serial algorithms, which are sequentially executed from the start to the end on one running thread. In real life, data is often very large, and serial algorithms can have long runtimes as they do not fully take advantage of the parallelism capabilities of modern multi-core CPUs. To address this limitation, this paper presents two parallel algorithms named DPCompact-SPADE ( D epth load balancing P arallel Compact-SPADE) and APCompact-SPADE ( A daptive P arallel Compact-SPADE) for weighted clickstream pattern mining. Experiments on various datasets show that the proposed parallel algorithm is efficient, and outperforms state-of-the-art serial algorithms in terms of runtime, memory consumption, and scalability.

Ontology verification refers to the activity where an ontology is tested against its ontology requirements to ensure that it is built correctly in compliance with its ontology requirements specification. Therefore, it is an important activity that should be performed in any ontology development process. Since manual verification can be a time-consuming and repetitive task, testing processes to automatically verify an ontology facilitate this activity. Moreover, the involvement of not only ontology engineers during the ontology verification process , but also domain experts and users, can provide valuable feedback to avoid misunderstandings and lack of information. This paper proposes a method for ontology verification that defines the testing activities to be performed. The method uses a testing language based on lexico-syntactic patterns to facilitate the definition of tests and an ontology to store and publish such tests. Moreover, this verification testing method proposes an online tool to execute tests on one or more ontologies. The method was compared in terms of time and errors by user evaluation with other tools for ontology verification; the evaluation showed that the tools that use testing languages had better results in terms of reducing errors in the verification activity compared to the tools that do not.

With the development of online social networks , user behavior prediction based on the data collected from these social networks has attracted increasing attention. In heterogeneous social networks, a node usually has several heterogeneous attributes to describe itself from different angles. However, most existing methods only utilize an attribute of each node and neglect other heterogeneous attributes. Therefore, this paper proposes a new user heterogeneous information embedding method, called user heterogeneous information embedding (UHIE). This method utilizes the attention mechanism to aggregate the heterogeneous attribute information of each neighbor to obtain their low-dimension representation. Then, the graph neural network is employed to aggregate the multi-relational information from neighbors to obtain the low-dimension representation of nodes. Furthermore, a new soft thresholding method is proposed to eliminate the unimportant information, called multi-head self-attention soft thresholding (MSST), which employs the multi-head self-attention mechanism to calculate an importance threshold for each feature. Based on UHIE and MSST, a new user behavior prediction model is proposed, called Heterogeneous Residual Self-Attention Shrinkage Network (HRSN). This model utilizes UHIE to aggregate heterogeneous information including all heterogeneous attribute information of nodes, and employs MSST to eliminate unimportant information. The experimental results on three real-world datasets show the superiority of the proposed model.

Named entity recognition and relation extraction are crucial tasks in natural language processing . As the traditional pipelined manners may suffer from the error propagation issue and ignore underlying interactions, joint extraction of entities and relations has become the dominant trend. However, the performance of existing joint extraction models needs improvement. This paper presents a two-stage tagging scheme that separately labels candidate head entities and multiple tail entities in specific relations. Next, it proposes a novel lightweight joint extraction neural model based on the entity-first labeling strategy. In the proposed model, the BiLSTM-based encoder combines the hidden state and global context features and feeds them as input for the next two entity labeling tasks. Further, with the input of the mixed context representation, the candidate-head-entity recognition module is adopted to identify the candidate head entity, while the multiple-tail-entities recognition module is equipped with an entity-correlated attention mechanism to identify the corresponding tail entity under a specific head entity. Comprehensive experiments on two widely used English datasets and one self-constructed Chinese dataset were performed. The experimental results showed that the proposed model outperformed the baseline approaches in the relation extraction task and achieved a competitive entity recognition effect via a lightweight architecture.

As an effective tool to deal with uncertain decision-making, three-way decisions (TWD) have gained wide attention in many applications. Decision-theoretic rough sets (DTRSs) as a classic model of TWD contain two key elements, i.e., conditional probability and loss functions. In this paper, we study the determination of these two elements in depth via the information interaction and modification strategy, and further propose a novel model of TWD. First, fuzzy c-means (FCM) is used to cluster the condition attribute information and loss function information, respectively. Considering the interaction of two types of information, we design the corresponding fusion tactic of these clustering results to get equivalent classes and develop a new calculation method of conditional probability. Then, we use probabilistic hesitant fuzzy sets (P-HFSs) to aggregate the loss functions of different members of the same equivalence class and get the probabilistic hesitant fuzzy elements (P-HFEs) loss functions. In this case, P-HFSs not only reflect the hesitant situation of decision-makers, but also depict the proportion of different opinions. Regarding P-HFEs loss functions, we also investigate the modification methods of its outliers in detail. Moreover, based on the P-HFEs loss functions, we propose TWD with probabilistic hesitant fuzzy decision-theoretic rough sets (P-HFDTRSs). Finally, in order to verify the effectiveness of our proposed method, we develop a series of comparative experiments and discuss the decision results on six UCI datasets.

Decoding is an important part of machine translation systems, and the most popular inference algorithm used here is beam search. Beam search algorithm improves translation by allowing a larger search space to be traversed than greedy search. However, as the beam width increases, the translation performance declines after a certain point in neural machine translation (NMT). This problem is usually not observed in statistical machine translation (SMT) due to the decoding method. This paper proposes a hybrid system based method that uses SMT predictions to prevent quality deterioration in the beam search algorithm used in NMT decoding. Our approach is based on the reranking n-best list of NMT according to the SMT system translation sentence. We propose two different algorithms for reranking NMT n-best lists. The first algorithm uses the length information of the SMT outputs. In contrast, the second uses a word-based similarity approach with the Jaccard Index, the Dice's Coefficient, and the Overlap Coefficient. Experiments on three different language pairs show that the method we propose prevents the decrease in translation quality and produces a gain of 1.3 BLEU and 1.6 METEOR for different beam sizes and 1.8 BLEU and 2.1 METEOR average scores compared to the baseline results. (c) 2021 Published by Elsevier Inc.

Due to small-quantity and often imbalance of labeled samples, it is challenging to establish a robust and accurate prediction model through data-driven methods. To deal with the small dataset problem, new virtual samples may be generated via virtual sample generation (VSG) methods based on the trend of the original small raw dataset, thereby improving modeling performance. Effective VSG is desirable, but also challenging. Conventional VSG usually assumes that the raw sample set contains only a single operating mode. Taking multi-mode into account will improve the VSG based modeling performance since actual processes are often multi-mode. To this end, an information expansion function considering sample density and amount (IEDA) is first developed to expand the domain range of the attributes in this paper. Then, virtual samples under the multiple operating mode condition are generated by proposing a Gaussian mixture model based virtual sample generation (GMMVSG) method. Applications of GMMVSG on Tennessee Eastman benchmark process and an industrial hydrocracking process show significant improvement of modeling and predictions over other conventional VSG methods.

This paper is concerned with the design of false data injection attacks (FDIAs) against fault detectors (FD) for linear cyber-physical systems (CPSs). The purpose of the attacker is to design an stealthy attack scheme such that the defenders cannot detect the faults in time or fail to detect the faults, and at the same time the robustness of the CPS is deteriorated. Unlike the existing optimal attack strategies (OASs) that are established based on sufficient conditions, necessary and sufficient conditions (NASCs) are established to maximize the degradation of the FD performance and robustness of CPS while maintaining stealth. Subsequently, an OAS is constructed by solving coupled backward recursive Riccati difference equations (RDEs). Finally, two simulation examples are employed to show the effectiveness of the designed attack scheme.

Multi-task learning is becoming more popular and is being applied in a variety of applications. It improves the accuracy of prediction by simultaneously learning related tasks and saves cost through shared structures. In particular, the prediction of event type from social data is also an area where multi-task learning can be utilized. In this paper, we present a novel deep learning framework called S patial E vents P rediction (SEP) based on multi-task learning to predict the types of events that happen at a specific location from social data. The proposed model focuses on predicting the attribute types of an event, which is referred to as subtypes. Specifically, an event type-specific attention mechanism is introduced to extract the representations of social data and to identify their important components. The proposed attention mechanism is based on a two-level attention, which measures the importance of words and sentences to the subtypes of an event. We also propose a representation sharing method using semantic and spatial relationships between locations to alleviate the sparsity and incompleteness of data. The proposed representation sharing preserves the spatial heterogeneity between locations and significantly improves the accuracy of the overall framework. Experiments with real-world datasets confirm the effectiveness and efficiency of the proposed method.

Fuzzy extractor can be used to convert the reading of a noisy non-uniform source into a reliably reproducible and almost uniform string, while reusable fuzzy extractor allows multiple extractions of the same source. In this paper, we propose multi-party fuzzy extractor, called threshold reusable fuzzy extractor, whose execution requires a threshold number of members. New definition and efficient construction are given with security proofs and experimental implementations. It preserves the inherent properties of fuzzy extractor with expanded use in multi-party scenarios. Based on the proposed threshold reusable fuzzy extractor, an application in privacy protection is given to store and share sensitive data by smart contract on blockchain , achieving joint access control via biometric information.

The recent success of deep convolutional neural networks (CNNs) is mostly due to the availability of large-scale datasets with accurate annotations. However, the collection of such large datasets with clean annotations is time-consuming and not always feasible. In this paper, we propose a novel framework for learning with noisy labels, called the Cyclical training method with Joint loss and Co-teaching strategy net (CJC-net), where the net means our method is insensitive to the structures of the CNN. CJC-net pretrains two networks simultaneously and then performs the cyclical training strategy under an improved co-teaching method based on the two pretrained networks. During the training process, we adjust the learning rates of the two networks to make the network states periodically transfer from overfitting to underfitting. The cumulative loss of each sample under two networks is recorded; and the higher the cumulative loss of a sample is, the higher the probability of identifying it as a noisy label or a hard label. Then, we remove those samples with a high loss and fine-tune the two networks using the remaining data. The experimental results on several datasets demonstrate that CJC-net is superior to many state-of-the-art methods.

The dynamics of rumor spreading in the information environment is a valuable way to study information management and public opinion monitoring. In this paper, the spreading of rumors and the behavior of crowd diffusion on network structure are studied by establishing a Suspicious-Infected ( SI ) reaction–diffusion model of rumor spreading with Allee effect. Especially, the necessary conditions for Turing pattern to appear in space are studied. Then, Monte Carlo simulation was used to prove the validity of the model. Furthermore, extensive numerical evaluation of the system is carried out under different networks, whose results show that the diffusion coefficient can change the patterns significantly. Then, on the lattice network and the cellular network , the spatial instability will cause the same pattern type, while the density distributions of S and I are no longer consistent on WS networks and BA networks. In addition, the change of the network structure and the introduction of the periodicity of diffusion coefficient will make the instability of the original system disappear in space. These works have certain guiding value to the prediction of rumor spreading scope.

Deep models have achieved prominent results in pattern recognition tasks, especially computer vision and natural language processing. However, the dataset bias caused by the distribution discrepancy between the training and testing data hinders the generalization ability of deep models. Though many domain adaptation approaches have been proposed to mitigate such negative effect, most of them improve the transferability of features by aligning global distributions of deep models. Few researchers pay attention to the versatility of deep features which can play a vital role in cross-domain recognition. In this paper, we propose to enrich the classic deep learning models by capturing high-low-frequency information and multi-scale features, which deal with the domain shift that cannot be easily addressed by merely feature-level alignment. The Hierarchical Transfer Network (HTN) leverages octave convolution, pyramid features, and self-attention mechanism for revamping the classic models, which can be further integrated with any domain alignment approaches by replacing the feature extractor with the proposed HTN. Extensive experiments have been conducted on three public domain adaptation benchmarks. The results show that the proposed HTN can effectively improve adversarial-based, statistics-based, and norm-based domain adaptation approaches, achieving competitive performance without involving model complexity.

The pattern-based anomaly detection method has obtained more attention since it was proposed. This is due to its ability to fully identify anomalies by considering two key features, including appearing rarely and deviating from most data elements. Although most pattern-based anomaly detection methods identify the anomalies from full data space, the process involved in performing this functionality is very time-consuming. Therefore, to solve this problem, this paper introduces a novel method, namely m inimal r are p attern-based anomaly detection method through considering a nti-monotonic c onstraints (MRPAC), for identifying anomalies in uncertain data. MRPAC detects the anomalies from a small scale of uncertain data that satisfy preset anti-monotonic constraints by mining constrained minimal rare patterns, thus, the time efficiency is increased. In addition, MRPAC also defines multiple deviation factors to compute the anomaly score for all transactions, to accurately discover potential anomalies through sorting their anomaly scores. Extensive experimental outcomes indicate that the MRPAC significantly outperforms five state-of-the-art pattern-based anomaly methods in terms of detection accuracy and time efficiency, and obtains good scalability.

Referring Expression Generation (REG) is one of the most relevant and complex tasks in Natural Language Generation . Plural referring expressions are linguistic expressions that intend to identify a set of objects in a certain context. In this paper we formulate for the first time plural referring expressions and the elements of the associated REG problem in terms of the well known Formal Concept Analysis (FCA) mathematical framework. We provide theoretical results and general algorithms for computing the whole collection of referable sets in the context under analysis, as well as a condensed representation of all referring expressions for each referable set.

A complex network is a network that has the characteristics of small world, clustering, and power-law distribution. The discovery of crucial spreaders, as one of the significant research directions in complex networks, is mainly used to identify nodes that play a key role in the structure and function of the network. The gravity model is a special method in identifying influencers. However, it involves an open issue that is how to determine the range of interaction. In addition, the mass is merely represented by degree of nodes in traditional methods, which is also a thought-provoking question. For the sake of solving the above two problems, this paper presents an effective gravity model which is based on the precise radius and value information. The rough truncation radius is accurately calculated. And the value information, which represents the dissemination ability of the node, is modified to mass. In short, the node’s influence range and value score are calculated according to the attributes of each node and the interaction of neighbor’s nodes in the network. Compared with other similar methods and state-of-the-art measures, the rationality and superiority of our approach are demonstrated through six experiments on eleven real-world networks.

In the field of evidence theory, the processing of uncertain information is a very important research orientation, which has been receiving widespread attention in academic community. Recently, Yager and Alajlan proposed a method to evaluate the belief structure satisfaction to uncertain target values. However, this method ignored the effect of the overlapping degree between events. In this paper, a new method used to calculate the belief structure satisfaction is proposed. With different overlapping degrees between events considered, this new method effectively quantifies the evaluation interval of the belief structure satisfaction. The validity of the proposed belief structure satisfaction is verified in an illustrative application for multi-criteria decision-making through a combination of evidential axiomatic design process.

Information overload in recent years has tremendously sparked recommender systems (RSs). An RSs usually recommends valuable information for users based on historical experience high-dimensional and incomplete (HDI) data. Since each user cannot mark whole items, extracting latent factors (LF) learned by the stochastic gradient descent (SGD) optimization method is frequently used. However, interference from the adjustment of parameters induces an inferior convergence rate and low efficiency. To address this issue, the paper proposes a b rain storm optimization (BSO)-based a daptative l atent f actor a nalysis (BALFA) model consisting of the following three essential ideas: 1) divergent mechanism to avoid premature convergence, 2) particle retention technique to prevent invalid search, and 3) self-adaptive multidimensional leaning rate for more efficient application on varying data. Moreover, the convergence of the modified BSO used in BALFA is proofed based on the Markov chain . Comparison experiments on six HDI datasets in the SGD-based LF model indicate that the BALFA achieves state-of-art convergence rate and computational efficiency for HDI data analysis.

In this paper, a novel self-adaptive fuzzy learning (SAFL) system is proposed for streaming data prediction. SAFL self-learns from data streams a predictive model composed of a set of prototype-based fuzzy rules, with each of which representing a certain local data distribution, and continuously self-evolves to follow the changing data patterns in non-stationary environments. Unlike conventional evolving fuzzy systems, both the fuzzy inference and consequent parameter learning schemes utilised by SAFL are simplified so that only a small number of selected fuzzy rules within the rule base are involved in system output generation and parameter updating during a learning cycle. Such simplification not only significantly reduces the system’s computational complexity but also increases its prediction precision. In addition, both theoretical and empirical investigations guarantee the stability of the resulting SAFL. Comparative experimental studies on a wide variety of benchmark and real-world problems demonstrate that SAFL is able to learn from streaming data in a highly efficient manner and to make predictions with a great accuracy, revealing the effectiveness and validity of the proposed approach.

Performance evaluation is significant in data classification. The existing evaluation methods ignore the characteristics (such as classification difficulty) of each instance. In practice, it is necessary to measure classification performance from the perspective of instances. In this paper, an instance-oriented classification performance metric is proposed based on the classification difficulty of each instance, named degree of credibility (Cr ). Cr conforms to the natural cognition that the lower the probability of misclassifying relatively easy instances, the more credible the classifier. It focuses on the credibility of each instance’s prediction, which opens up a new way for classifier evaluation. Moreover, several important properties of Cr are identified, laying solid theoretical foundation for classifier evaluation. Also, the concept of acceptable classifier is proposed to judge whether the trained model and its parameter set reach excellent ranks at the current technology level instead of relying entirely on human experience. The experimental results of twelve classifiers on twelve datasets indicate the physical significance and good statistical consistency and discriminatory ability of Cr, as well as the feasibility of acceptable classifiers for model selection and training. Furthermore, the proposal of approximate difficulty greatly improves the computation efficiency of instance difficulty.

This paper presents a new algorithm called general type-2 collaborative fuzzy clustering (GT2-CFC), which focuses on uncertainty associated with the structures discovered for the same data stored in multiple data sites in different feature spaces while maintaining data-sharing restrictions. The uncertainty of membership degrees is directly controlled by using the available data rather than by reinforcing the lower-order fuzzy model through general type-2 fuzzy sets (GT2FS) including primary and secondary memberships. Primary membership degrees show the obtained membership without the knowledge gained from other data sites, and secondary membership degrees indicate the belonging degree to the data in the presence of external knowledge. In addition to the ability of GT2FS in uncertainty modeling, due to the lack of any type reduction and defuzzification during the proposed clustering process, there is no data loss, and time complexity becomes is reduced leading to more powerful uncertainty modeling. Finally, a new validity index is proposed to evaluate the quality of collaboration using the relative entropy-based global consistency. The superiority and efficiency of the proposed method are illustrated with several experiments.

Active learning (AL) for object detection (OD) aims to reduce labeling costs by selecting the most valuable samples that enhance the detection network from the unlabeled pool. Due to the complexity of OD compared with image classification , more consideration should be given when designing the selection strategies. Previous works have studied aggregating information of multiple outputs (especially the location information) and aggregating information of batch boxes, all of which indicate improved performances. However, the evaluation index—mean average precision (mAP) has not been considered seriously, although improving it is the goal of AL. Moreover, the background class is far more than other classes (15:1 or more) in each batch of samples, leading to a class imbalance problem . Therefore, AL strategies for OD, which take mAP and class imbalance in batch into consideration, may perform better. In this paper, WBetGS is proposed, which not only considers aggregating information of multiple outputs and batch boxes but also aims to mAP improvement and to address the class imbalance in batch. A weighted algorithm is introduced to promote the mAP more effectively. Besides, WBetGS eliminates the impact of class imbalance between background and object categories by extracting class-balanced information. Moreover, a diversity and uncertainty based sampling algorithm is introduced for batch mode active learning in object detection. The experimental results demonstrate that our method performs better than basic methods, saving up 100% of the labeling efforts while reaching the same performance in an actual industrial application.

The massive growth of audiences eager for sport content has substantially increased workers’ demand in this profitable segment. Highlight identification is vital for summarizing football matches. Decision support tools can significantly reduce the number of company employees required to tackle such a task, widely benefiting workforce resource allocation. This paper discusses the development of an automatic football highlight detector. The proposed system exploits discriminative low-level audio and video features extracted from a compact set of irregularly time–spaced frames that integrate a long-term sliding window. A new mixed wrapper-probabilistic algorithm leverages a cost-effective selection of the most significant frames submitted to a robust multi-frame consensus classification scheme . By considering a comprehensive database integrating 30 full matches, the proposed approach achieves a highlight identification rate of 100% (including all annotated goals), conjugated with a match-time compression rate of about 94%, when employing a Random Forest classifier .

Domain adaptation has attracted attention by leveraging knowledge from well-labeled source data to facilitate unlabeled target learning tasks. Numerous research efforts have been devoted to extracting effective features by incorporating the pseudolabels of target data. However, the transferable knowledge reflected by intradomain structure, interdomain correlation and label supervision has scarcely been considered simultaneously. In this paper, we propose a novel structure preservation adversarial network with target reweighting (SPTR) for unsupervised domain adaptation, in which local structure consistencies and category-level semantic alignment are simultaneously considered in the adversarial learning framework. Based on the labeled and pseudolabeled samples, we attempt to align both global and category-level domain statistics from different domains and simultaneously enforce structural consistency from feature space to label space in the source and target domains. Furthermore, to suppress the influence of falsely labeled target samples, a novel and generalized sample reweighting strategy is developed to assign target samples with different levels of confidence, which fully explores the knowledge of the target distribution to benefit the semantic transfer process. The experimental results in three transfer learning scenarios demonstrate the superiority of our proposed method over other state-of-the-art domain adaptation algorithms.