Search

The ever-increasing size of images has made automatic image annotation one of the most important tasks in the fields of machine learning and computer vision. Despite continuous efforts in inventing new annotation algorithms and new models, results of the state-of-the-art image annotation methods are often unsatisfactory. In this paper, to further improve annotation refinement performance, a novel approach based on weighted mutual information to automatically refine the original annotations of images is proposed. Unlike the traditional refinement model using only visual feature, the proposed model use semantic embedding to properly map labels and visual features to a meaningful semantic space. To accurately measure the relevance between the particular image and its original annotations, the proposed model utilize all available information including image-to-image, label-to-label and image-to-label. Experimental results conducted on three typical datasets show not only the validity of the refinement, but also the superiority of the proposed algorithm over existing ones. The improvement largely benefits from our proposed mutual information method and utilizing all available information.

This study shows the influence of human resource management practices on the academic and non-academic staff and its impact on the higher education institution's goals. A questionnaire based on the Cranet survey was used to gather data from 240 employees (academic and non-academic staff) from a public higher education institution. A phi-k correlation algorithm was used to verify the underlying correlation coefficients, statistical significance, and outliers within multiple data types. This algorithm allows a more personalized, understandable approach to reveal the human resource management practices that significantly impact the teacher career development and capacity improvement. In addition, the use of background variables to identify the groups of respondents allows the algorithm to discern the multidimensional data for a more personalized human resource management approach. Human resource management practices involving training development and staff were correlated to the institution's goals. The phi-k correlation proved to be a suitable tool to shape structural models and latent trait analysis between multiple data types, which can overcome the drawbacks of Pearson and Cramer correlations when processing non-linear data. The presented research contributes to the literature by using the phi-k algorithm to process multiple data types. The proposed study with the phi-k algorithm is the first time applied to higher education institutions.

The capacity of combined gauge-satellite precipitation estimates largely depends on the characteristics of the input data such as the number, location and reliability of rain gauges, and satellite-derived precipitation quality. The objective of this study is to examine the influence of rain gauge network configuration including density and spatial distribution on the performance of the gauge-satellite merging estimation at monthly and ten-day temporal scales. Dense rain gauge observations and satellite-derived precipitation data (i.e., TMPA 3B42 Version 7 and Version 06 IMERG Final Run) in two provinces of China are used. A two-stage downscaling-integration approach is applied in the gauge-satellite precipitation estimation. Various scenarios of rain gauge density and combination are designed and their corresponding merged precipitation estimates are evaluated using statistical indices. The merged results using the TMPA and IMERG precipitation product, respectively, are compared. The results show that: 1) the influence of rain gauge network configuration on the gauge-satellite merged precipitation estimates gradually decreases with the increase in rain gauge density, and the gauge-satellite merged precipitation estimates are more sensitive to the rain gauge network density in wet season and ten-day temporal scale than in dry season and monthly scale, respectively and 2) the merged precipitation estimation using the IMERG precipitation data generally outperforms the estimation using TMPA precipitation data in the low gauge density scenarios, and the gap decreases with the increase in the rain gauge network density. In the areas with sparse rain gauges, improving the quality of satellite precipitation data would significantly improve the performance of the gauge-satellite merging estimation.

Background: The recurrence and liver metastasis rates are still high in pancreatic head cancer with curative surgical resection. A no-touch isolation principle in pancreaticoduodenectomy (PD) may improve this situation, however, the exact advantages and efficacy of these principles have not been confirmed. Materials and methods: Among 370 patients who underwent PD, three centers were selected and classified into two groups: the no-touch PD group (n = 70) and the conventional PD group (n = 300). Propensity score matching was used to control for selection bias at a ratio of 1:1. The confounding variables were age, sex, body mass index, adjuvant chemotherapy, carbohydrate antigen 19-9, tumor size and tumor differentiation. Results: Patients in the no-touch PD group had better overall survival (OS) and disease-free survival (DFS) than those in the conventional PD group (OS: 17 vs. 13 months, p = 0.0035, DFS: 15 vs. 12 months, p = 0.087), with lower 1- and 2-year disease-related mortality rates (1-year: 32.9% vs. 47%, p = 0.032; 2-year: 42.5% vs. 82% p = 0.000) and recurrence and liver metastasis rates (1-year: 30.0% vs. 43.3%, p = 0.041; 2-year: 34.3% vs. 48.7%, p = 0.030). Compared with the matched conventional PD group, the no-touch PD group also had a better OS (17 vs. 12 months, p = 0.032). Conclusions: Our study showed the no-touch isolation principle may be a better choice to improve long-term survival for pancreatic cancer patients.

This article addresses the visual regulation problem of a differential-drive mobile robot with an arbitrarily installed monocular camera in the indoor environment. A three-stage controller is designed by using a novel nonadaptive switching approach, where the unknown image depth and the uncalibrated camera-to-robot translation parameters do not need to be estimated. Convergence of the error systems with the designed controller in each stage is analyzed. Moreover, the existence of the switching time instants from each stage is proved. The simulation results are presented to show the effectiveness of the proposed approach.

Automatic image annotation is a key technology in image understanding and pattern recognition, and is becoming increasingly important in order to annotate large-scale images. In the past decade, the nearest neighbor model-based AIA (Automatic image annotation) methods have been proved to be the most successful in all classical models. This model has four major challenges including semantic gap, label-imbalance, wider range labels, and weak-labeling. In this paper, we propose a novel annotation model based on three-pass KNN (k-Nearest Neighbor) to address the aforementioned challenges. The key idea is to identify appropriate neighbors at each pass KNN. In the first pass KNN, we identify the several most relevant categories based on label feature rather than visual feature as traditional models. In the second pass KNN, we determine the relevant images based on multi-modal (visual and textual label) embedding features. As the test image has not been annotated with any label, we propose a pre-annotation strategory before image annotation to improve the semantic level. In the third pass KNN, we capture relevant labels from semantically and visually similar images and propagate them to the given unlabeled image. In contrast with traditional nearest neighbor based methods, our method can inherently alleviate the problems of semantic gap, label-imbalance, and wider range labels. In addition, to alleviate the issue of weak-labeling, we propose label refinement for training images. Extensive experiments on three classical benchmark datasets and MS-COCO demonstrate that the proposed method significantly outperforms the state-of-the-art in terms of per-label and per-image metrics.

The traffic scene understanding is the core technology in Intelligent Transportation Systems (ITS) and Advanced Driver Assistance System (ADAS), and it is becoming increasingly important for smart or autonomous vehicles. The recent methods for traffic scene understanding, such as Traffic Sign Recognition (TSR), Pedestrian Detection, and Vehicle Detection, have three major shortcomings. First, most models are customized for recognizing a specific category of traffic target instead of general traffic targets. Second, as for these recognition modules, the task of traffic scene understanding is to recognize objects rather than make driving suggestions or strategies. Third, numerous independent recognition modules disadvantage to fusing multi-modal information to make a comprehensive decision for driving operation in accordance with complicated traffic scenes. In this paper, we first introduce the image captioning model to alleviate the aforementioned shortcomings. Different from existing methods, our primary idea is to accurately identify all categories of traffic objects and understand traffic scenes by making full use of all information, and making the suggestions or strategy for driving operation in natural language by using Long Short Term Memory network (LSTM) rather than keywords. The proposed solution naturally solves the problems of feature fusion, general object recognition, and low-level semantic understanding. We tested the solution on our created traffic scene image dataset for evaluation of image captioning. Extensive experiments including quantitative and qualitative comparisons demonstrate that the proposed solution can identify more objects and produce higher-level semantic information than the state-of-the-arts.

Laparoscopic cholecystectomy is the standard treatment for cholelithiasis. A very rare condition named situs inversus should not be considered as a contraindication for laparoscopic cholecystectomy. Here, we reported a case of successful laparoscopic cholecystectomy in a patient with situs inversus totalis. We also described the technical advantages of this treatment and reviewed the literature.

PurposeThe Contingency Outsourcing Relationship (CORE) model originated from the Four Outsourcing Relationship Types (FORT) model and the CORE model is used in the globalized facility management (FM) industry while the FORT model is originally used in the global information technology industry. The purpose of this paper is to analyze the CORE model through the rankings of relationship between a client and a globalized FM service provider from the perspective of the FM service provider in one of the four categories (i.e. in-house, technical expertise, commitment and common goals) and the application of this model with the aid of artificial neural networks (ANNs).Design/methodology/approachA quantitative methodology using a survey is used to analyze the four types of outsourcing categories. First, the background theory and a set of rules of the CORE is introduced and discussed regarding the proper ways to identify the rankings collected from the survey.FindingsThe study reveals that an interesting understanding of the outsourcing categories can be systematically implemented into the FM outsourcing relationships through the methodology of scientific artificial intelligence. FM outsourcing categorization may help to define the appropriate relationship; as either not too aggressive or too passive.Originality/valueThe outcome generated from the ANN can be considered a strong and solid reference to assess and define the existing outsourcing relationships between the stakeholders and the service providers with the goal to assign an outsourcing category to the service provider based on the learnt rules.

This paper is concerned with the distributed estimation problem in sensor networks subjected to unknown attacks. Network attacks are considered to exist in two classes of channels: 1) communication channels from the plant to sensors and 2) communication channels among sensors. The status of an attack is viewed as a stochastic phenomenon, and the transmitted information will be affected when the attacker successfully carries out an attack on the related data packet. Based on the sensors’ own measurements and their neighbors’ local information, a novel distributed estimation model against two-channel stochastic attacks is presented. A sufficient condition on the existence of the desired distributed ${H} _{\boldsymbol {\infty }}$ estimators is derived and the distributed estimator gains are designed by solving a linear matrix inequality. Two illustrative examples are provided to demonstrate the effectiveness of the new design techniques.

Automatic image annotation plays a significant role in image understanding, retrieval, classification, and indexing. Today, it is becoming increasingly important in order to annotate large-scale social media images from content-sharing websites and social networks. These social images are usually annotated by user-provided low-quality tags. The topic model is considered as a promising method to describe these weak-labeling images by learning latent representations of training samples. The recent annotation methods based on topic models have two shortcomings. First, they are difficult to scale to a large-scale image dataset. Second, they can not be used to online image repository because of continuous addition of new images and new tags. In this paper, we propose a novel annotation method based on topic model, namely local learning-based probabilistic latent semantic analysis (LL-PLSA), to solve the above problems. The key idea is to train a weighted topic model for a given test image on its semantic neighborhood consisting of a fixed number of semantically and visually similar images. This method can scale to a large-scale image database, as training samples involved in modeling are a few nearest neighbors rather than the entire database. Moreover, this proposed topic model, online customized for the test image, naturally addresses the issue of continuous addition of new images and new tags in a database. Extensive experiments on three benchmark datasets demonstrate that the proposed method significantly outperforms the state-of-the-art especially in terms of overall metrics.

Nowadays, with the rapid growth of imaging and social network, huge volumes of image data are produced and shared on social media. Social image annotation has been an important and challenging task in the fields of computer vision and machine learning, which can facilitate large-scale image retrieval, indexing, and management. The four most challenges of social image annotation are semantic gap, tag refinement, label-imbalance, and annotation efficiency. To address these issues, we propose an efficient and effective annotation method based on the Mean of Positive Examples (MPE) corresponding to each label. First, we refine user-provided noisy tags by our proposed local smoothing process, and consider the refined tags as key features in contrast to the previous methods that consider them as side information, which significantly improves annotation performance. Second, we propose a weighted trans-media similarity measure method that fuses all modality information in identifying proper neighbors, which promotes the semantic level and eases image annotation. Third, our MPE model gives equal importance to all labels, thus, improving the annotation performance of infrequent labels without sacrificing that of frequent labels. Fourth, our MPE model can dramatically decrease space-time overheads, since the time cost of annotating an image is unaffected by the size of the training image dataset, but relying on the size of label vocabulary. Therefore, our proposed method can be applied to real-world large-scale online social image repositories. Extensive experiments on both benchmark datasets demonstrate the effectiveness and efficiency of our MPE model.

The consensus tracking problem means that a group of followers tracks the desired trajectory with local communication. In this article, partial components of cluster consensus have been considered. In this scenario, the p components of the followers in different clusters track the leader at different lag times, while p components of each agent in the same cluster reach a consensus, which is called p components of cluster-lag (PCCL) consensus. By using a seminorm $||x_{i}||_{2,p}$ and a Lyapunov-Krasovskii functional, PCCL consensus for second-order multiagent systems with homogeneous nonlinear systems on cooperative-competitive networks has been considered. For the case that the communication network graph is undirected, a decentralized adaptive controller, which is based on the exchanged neighbors' information from the same cluster, is designed such that all the agents reach PCCL consensus. For the directed graph case, an adaptive protocol based on the intracoupling strength is constructed for each cluster to achieve PCCL consensus. Finally, two simulation examples are illustrated to show the effectiveness of the proposed control protocols.

The ambiguity of name entities is a common problem in information retrieval, which leads to the decline of retrieval quality. This makes name disambiguation particularly important. In academic field, the rapidly increasing large-scale of publications has imposed more challenges to the name disambiguation problem. Existing works mainly focus on leveraging content information to distinguish different name entities. In this paper, we consider jointly utilizing both content information and relational information to disambiguate the same name. Firstly, we construct a Heterogeneous Academic Network based on meta information of publications such as collaborators, institutions and venues. Then, we transform the network into separate homogeneous graphs. After that, we propose Graph Attention Networks to jointly learn content and relational information by optimizing an embedding vector. Finally, a clustering algorithm is presented to gather author names most likely representing the same person. The experiments show that our method is effective and outperforms the state-of-the-art methods in both precision and recall metrics.

In this paper, the global pinning synchronization problem of a class of complex dynamical networks with hybrid couplings, time delays, random data packet dropouts and multiple stochastic disturbances is investigated. The hybrid couplings are portrayed in three forms: constant couplings, discrete-delay couplings and distributed-delay couplings. The phenomenon of random packet dropouts is described as a binary random variable which obeys the Bernoulli distribution taking values of 0 and 1 with a certain probability. Multiple noisy processes herein are characterized by Brownian motions, which act on all coupling terms as well as the overall dynamics equation. By applying the Lyapunov method and stochastic analysis technique, sufficient conditions are established to ensure the considered stochastic delayed complex network to globally synchronize to a reference trajectory in the mean-square sense via the developed randomly occurring pinning control strategy. The obtained criteria are within the framework of linear matrix inequations. Finally, a simulation example is presented to verify the feasibility and effectiveness of the derived theoretical results.

Aspect-based Sentiment Analysis (ABSA) is a subclass of sentiment analysis, which aims to identify the sentiment polarity such as positive, negative, or neutral for specific aspects or attributes that appear in a sentence. Previous studies have focused on extracting aspect-sentiment polarity pairs based on dependency trees, ignoring edge labels and phrase information. In this paper, we instead propose a phrase dependency graph attention network (PD-RGAT) on the ABSA task, which is a relational graph attention network constructed based on the phrase dependency graph, aggregating directed dependency edges and phrase information. We perform experiments with two pre-training models, GloVe and BERT. Experimental results on three benchmarking datasets (i.e., Twitter, Restaurant, and Laptop) demonstrate that our proposed PD-RGAT has comparable effectiveness to a range of state-of-the-art models and further illustrate that the graph convolutional structure based on the phrase dependency graph can capture both syntactic information and short long-range word dependencies. It also shows that incorporating directed edge labels and phrase information can enhance the analysis of aspect-sentiment polarities on the ABSA task.

In this paper, a fusion estimation problem in clustering sensor networks under stochastic deception attacks is investigated. The sensors are divided into several clusters, and a local estimator is ...

In this paper, we consider a periodic estimation problem in sensor networks with a shared communication channel. The transmission constraint is inevitable in a single-channel-based sensor network if the sensors are heterogeneous or deployed far away from each other. A novel stochastic competitive transmission strategy is presented to deal with the transmission constraint, such that the sensors communicate with the fusion center (FC) in a strict asynchronous manner. A periodic mixed storage strategy combing the zero-input and the hold-input mechanisms is presented to describe periodic updating of the stored information in the sensors' buffers. A recursive Kalman filtering algorithm is derived for the FC to periodically generate estimates of state variables describing an object by using a linear continuous-time stochastic model. Two simulation examples are presented to show the effectiveness of the proposed results.

This paper is concerned with the state estimation problem for a class of nonlinear cyber-physical systems (CPSs) where the nonlinear dynamical physical process is approximated by a Takagi–Sugeno fuzzy model. The physical plant is measured by a set of wireless sensors and the sensors communicate with the remote estimator via a communication channel. In the considered CPS, the randomly occurring sensor saturation, signal quantization, packet dropouts as well as the medium access constraint are studied in a unified framework. We develop a sufficient condition such that the filtering error system is asymptotically stable in the mean-square sense and also with a prescribed $\boldsymbol {H_\infty }$ performance level. The filter gain parameters are determined by solving a convex optimization problem. Finally, the simulation study on the networked truck-trailer system is presented to show the effectiveness of the proposed estimator design.

This paper is concerned with the set-membership estimation problem for complex networks subject to unknown but bounded attacks. Adversaries are assumed to exist in the nonsecure communication channels from the nodes to the estimators. The transmitted measurements may be modified by an attack function with added noise that is determined by the adversary but unknown to the estimators. A novel set-membership estimation model against unknown but bounded attacks is presented. Two sufficient conditions are derived to guarantee the existence of the set-membership estimators for the cases that the attack functions are linear and nonlinear, respectively. Two strategies for the design of the set-membership estimator gains are presented. The effectiveness of the proposed estimator design method is verified by two simulation examples.

Pose estimation problem is concerned with determining position and orientation of an object in real time using the image information, and has found applications in many fields such as object recognition and robotic visual servoing. Most of vision-based pose estimation schemes are derived from extended Kalman filter, which requires that the noises obey the Gaussian distribution under known covariance. However, the statistical information in robot control may not be accurately obtained or satisfied. In this paper, a novel bounded recursive optimization approach is proposed to solve the pose estimation problem in visual serving, where the addressed noises do not provide any statistical information, and the bounds of noises are also unknown. Finally, the pose estimation simulation is conducted to show the advantages and effectiveness of the proposed approach.

Spindle microtubules (MTs) capture kinetochores (KTs) on the centromere sequence of sister chromatids to align at the mitotic equator and segregate toward spindle poles during mitosis. For efficient chromosome capture, KTs initially attach to the lateral surface of a MT, providing a considerably larger contact surface than the MT tip. A sequential change of KT composition upon spindle attachment enables a conversion from lateral to stable end-on attachment. However, the molecular link between spindle dynamics and KT composition is not fully understood. Here, we report that Ska1 and DDA3 act as molecular linkers in the interplay between KTs and spindle dynamics. After recruitment of Kif2a onto the mitotic spindle by DDA3, Ska1 targets Kif2a to the minus-end of spindle MTs and facilitates spindle dynamics. Furthermore, DDA3 targets Ska1 to KTs to stabilize end-on attachment. Thus, our findings identified a definite regulatory mechanism of the search and capture process for stable spindle attachment through cross-talk between spindle dynamics and KT composition mediated by DDA3 and Ska1.

This paper is concerned with the distributed estimation problem in sensor networks subjected to unknown attacks. Network attacks are considered to exist in two classes of channels: 1) communication channels from the plant to sensors and 2) communication channels among sensors. The status of an attack is viewed as a stochastic phenomenon, and the transmitted information will be affected when the attacker successfully carries out an attack on the related data packet. Based on the sensors' own measurements and their neighbors' local information, a novel distributed estimation model against two-channel stochastic attacks is presented. A sufficient condition on the existence of the desired distributed H

This paper investigates the least-squares estimation problem for networked systems with transmission constraints. A group of sensors are deployed to measure the outputs of a plant and send the measurements to an estimator through a common communication channel. Due to the transmission constraints caused by the heterogenous or long-distance deployed sensors, only one sensor is allowed to transmit its measurement over one time slot. In this regard, a stochastic competitive transmission strategy is proposed to schedule the transmission permissions. By using the least-squares estimation approach, an aperiodic multi-step estimation algorithm is proposed for the estimator to aperiodically generate the estimates. Performance analysis is presented for the estimation system with bounded noises and random noises. An upper bound is derived for the expectation of the estimation error and a sufficient condition is presented to ensure the convergence of the obtained upper bound. An illustrative example is provided to demonstrate the effectiveness of the proposed results.

This technical note is concerned with the distributed Kalman filtering problem for a class of networked multi-sensor fusion systems (NMFSs) with communication bandwidth constraints. To satisfy finite communication bandwidth, only partial components of the local vector estimation signals are transmitted to the fusion center (FC) at each time step, where multiple binary variables are introduced to model this component transmitting process. A novel compensation strategy is proposed to restructure the untransmitted components of each local estimate at the FC end, and a recursive distributed fusion kalman filter (DFKF) is designed in the linear minimum variance sense. Moreover, a simple suboptimal judgement criterion is proposed to determine a group of binary variables such that the mean square error of the designed DFKF is minimal at each time step. An illustrative example is given to show the effectiveness of the proposed methods.

The Event Service defined in Common Object Request Broker Architecture provides an asynchronous, mul- ticast communication model among distributed objects. However, the standard CORBA Event Service and previous methods lack important features required by real-time applications. For instance, message transferring programs for coop- erating design groups may have requirements of real-time processing and persistent storage of Event data. To address the- se problems, we propose a real-time message middleware design based on Event service. First, we extend Standard Event Service to a real-time Message Middleware by improving the QoS (Quality of Service) of Event Channel. Second, we propose a model that can persistently store the Event data and recover after the system crushed. Finally, we introduce an Event Channel Manager object to well manage the Event Channels. By using this real-time message middleware, the Ob- ject Request Brokers can communicate stably with each other without caring whether the two or many communication sides have relation or whether the other communication side is ready.

This letter investigates the hierarchical fusion estimation for clustered sensor networks. The sensors within the same cluster are connected to a local estimator, and all the local estimators are linked with a fusion center. The fusion center and the local estimators are not required to be synchronous. During each estimation interval, the sensors are allowed to communicate with the local estimator several times. A minimum variance estimation algorithm is presented for each cluster to aperiodically generate local estimates. A covariance intersection fusion strategy is presented for the fusion center to generate fused estimates by using asynchronous local estimates and previous fused estimates, without knowing the cross-covariances among the local estimation errors.

Motion blur appearing in traffic sign images may lead to poor recognition results, and therefore it is of great significance to study how to deblur the images. In this paper, a novel method for deblurring traffic sign is proposed based on exemplars and several related approaches are also made. First, an exemplar dataset construction method is proposed based on multiple-size partition strategy to lower calculation cost of exemplar matching. Second, a matching criterion based on gradient information and entropy correlation coefficient is also proposed to enhance the matching accuracy. Third, L0.5-norm is introduced as the regularization item to maintain the sparsity of blur kernel. Experiments verify the superiority of the proposed approaches and extensive evaluations against state-of-the-art methods demonstrate the effectiveness of the proposed algorithm.

This study investigates the signal estimation problem in noisy sensor networks with quantised communications. The sensors are subject to random sensor failures, and synchronously take noisy measurements to produce local estimates by using a Kalman filtering scheme at each sampling instant. A quantiser is considered to be embedded in each sensor, and the probabilistic quantisation strategy is adopted to reduce the energy consumption. In between two sampling instants, each sensor collects quantised local estimates from its neighbours and runs a consensus-based fusion algorithm to generate a fused estimate. The process noises and measurement noises are considered to be spatially uncorrelated, a recursive equation is presented to calculate the estimation error covariance matrix and an upper bound is derived for the estimation performance index. Moreover, a sufficient condition for the convergence of the upper bound of the estimation performance index is also presented. Two types of optimisation problems are constructed for cases of infinite and finite recursions, respectively, where the former one focuses on minimising the derived upper bound of the estimation performance index, and the latter one aims to minimise the energy consumption subject to a constraint on the estimation performance. Illustrative examples are provided to demonstrate the effectiveness of the proposed theoretical results.

This study investigates the multi-sensor-based centralised estimation problem in heterogeneous sensor networks with a common communication channel. Owing to the heterogeneity of the distributed sensors, it is usually impossible to package the measurements into one packet and transmit them to the fusion centre (FC) together, which implies that the measurements should be forwarded to the FC asynchronously. In view of this, a novel stochastic competitive transmission strategy is proposed to stagger the sensors’ transmissions. By using the asynchronous sampled information from the sensors, an H ∞ filter is designed for the FC to periodically generate estimates. The filter parameters are determined by solving a linear-matrix inequality. An illustrative example is provided to demonstrate the effectiveness of the proposed theoretical results.

Research on real-time 3D animation is attracting a lot of attention in recent years due to the popularity of emerging applications such as distributed virtual environments and computer games. One of the important issues in real-time animation is that the existing motion retrieval techniques generally have a high matching time because they are typically based on matching time-series, making them less suitable for use with large motion databases. In this paper, we propose a different approach to motion retrieval, called Eigen-based Motion Retrieval (or EigenMR), to address this limitation of the existing methods by performing motion retrieval in the transform domain instead of the time domain. To differentiate the motion of different body parts, we propose to perform the matching on individual body parts as well as on the whole body. Our approach has the important advantage that each body part can be represented by an index of fixed size, consisting of a number of eigenvectors and the corresponding eigenvalues. As a result, our approach has constant time complexity based on the number of motion files in the database instead of the size of the database. The experimental results show that our approach is both efficient and accurate compared with some of the latest methods. When applied to a motion database of 4GB in size, our method requires approximately 20% of the standard time, making it more suitable for real-time animation.

This paper is concerned with the fuzzy-model-based fault detection for a class of nonlinear systems with networked measurements where there are significant uncertainties on information. A unified model is proposed to capture four sources of these uncertainties, namely, the sensor saturation, the signal quantization, the general medium access constraint, and the multiple packet dropouts. A simultaneous consideration of these issues reflects the practical networked systems much more closely than the existing works. The goal of this paper is to design a fault detector such that, for all unknown input, control input, and uncertain information, the estimation error between the residual and the fault is minimized. Using the switched system approach and some stochastic analyses, a sufficient condition for the existence of desired fault detector is established and the fault detector gains are computed by solving an optimization problem. Two numerical examples are given to show the effectiveness of the proposed design.

This paper is concerned with a distributed set-valued estimation problem in noisy sensor networks with limited communication data rate. The sensors iterate a sample-quantize-exchange process: firstly, the sensors synchronously take noisy measurements of the plant; secondly, owing to the communication channel constraint, each sensor makes the manipulation of quantization on its measurement, estimate and performance bound by using a static logarithmic quantizer; thirdly, the quantized information is exchanged among the neighbors. By assuming hard bounds on the process noises and measurement noises, a distributed set-valued estimation strategy is proposed for each sensor to generate a set of estimates in state space which contains the true state of the plant. A sufficient condition for the existence of the distributed set-valued estimators is derived, and the estimator parameters are determined by solving a convex optimization problem. Two illustrative examples are provided to demonstrate the effectiveness of the proposed theoretical results.

Motion capture data in an uncompressed form can be expensive to store, and slow to load and transmit. Current compression methods for motion capture data are primarily lossy and cause distortions in the motion data. In this paper, we present a lossless compression algorithm for motion capture data. First, we propose a novel Alpha Parallelogram Predictor (APP) to estimate the DOF (degree of freedom) of each child joint from those of its immediate neighbors and parents that have already been processed. The prediction parameter of the predictor, which is referred to as the alpha parameter, is adaptively chosen from a carefully designed lookup table. Second, we divide the predicted and actual values into three components: sign, exponent and mantissa. We then compress their corrections separately with context-based arithmetic coding. Compared with other lossless compression methods, our approach can achieve a higher compression ratio with a comparable compression time. It can be used in situations where lossy compression is not preferred.

This paper investigates a group consensus problem with discontinuous information transmissions among different groups of dynamic agents. In the group consensus problem, the agents reach more than one consistent state asymptotically. We consider that the communication topology of these agents, represented by a network, is undirected. Then a novel group consensus protocol, called hybrid protocol, is proposed to solve the couple-group average-consensus problem. The convergence analysis is presented and the algebraic criterions are established. Furthermore, the multi-group consensus is discussed as an extension of the couple-group consensus. By similar techniques, some analysis results are presented. The analysis tools developed in this paper are based on algebraic graph theory, matrix theory, and control theory. Finally, the simulations are provided to demonstrate the effectiveness of the proposed theoretical results.

This paper is concerned with the multi-sensor-based estimation problem in wireless sensor network. A stochastic competitive transmission strategy is proposed to stagger the sensors' transmissions. By using the asynchronous sampled information from the sensors, a recursive Kalman estimation algorithm is derived for the fusion center to periodically estimate several future states of the target. An illustrative example is provided to demonstrate the effectiveness of the proposed theoretical results.

This paper is concerned with the finite-time quantized H∞ control problem for a class of discrete-time switched time-delay systems with time-varying exogenous disturbances. By using the sector bound approach and the average dwell time method, sufficient conditions are derived for the switched system to be finite-time bounded and ensure a prescribed H∞ disturbance attenuation level, and a mode-dependent quantized state feedback controller is designed by solving an optimization problem. Two illustrative examples are provided to demonstrate the effectiveness of the proposed theoretical results.

In Chaps. 2, 3, and 4, energy-efficient fusion estimation methods are presented by slowing down the transmission rates of measurements/local estimates and the estimation rate. In this chapter, a dimension-reduction method will be introduced for energy-efficient fusion estimation. To satisfy finite communication bandwidth and save energies consumed in communication, different dimensionality reduction approaches have been proposed in [1–7] to solve the fusion estimation problem, and the main idea of these approaches is that all the components of a vector signal are weighted and added to realize the objective of dimension reduction. Note that one should resort to the feedback information from a fusion center to obtain the compression matrices [3]. Different from the existing methods, this chapter presents the idea of directly choosing a part of components of local estimates to reduce the dimension of the local estimates to be transmitted to a fusion center. Specifically, when a local estimate is available at each sensor, only a part of the elements of the local estimate is selected and transmitted to the fusion center to save energy and meet the network bandwidth constraint. After the fusion center receives the local estimate with reduced dimension, a compensation strategy is proposed to reconstruct the local estimate and design the local unbiased estimator and improve the fusion estimation precision. Based on the optimal fusion estimation algorithm weighted by matrices, a recursive distributed fusion estimator is designed in the linear minimum variance sense. The gain matrix of the designed fusion estimator can be computed off-line as it does not need to know whether each component is sent or not at a particular time. Since the performance of the fusion estimator is dependent on the local estimate components selecting probabilities, some sufficient conditions, which are related to the selecting probabilities and system parameters, are derived such that the mean square error (MSE) of the fusion estimator is bounded. For linear time-invariant systems, some sufficient conditions are presented for the convergence of the fusion estimators.

By quantization, one is able to reduce the size of data packet containing the quantized signal and thus is able to satisfy the bandwidth constraint of the sensor network and reduce communication costs from the sensors to the fusion estimator. In this chapter, a design method for the \(H_{\infty }\) multisensor fusion estimator will be presented for sensor networks with quantized local estimates. The \(H_{\infty }\) estimator does not make any assumption on the statistics of the process and measurement noises; the only assumption is that the external disturbance has bounded energy [1, 2]. A group of finite-level logarithmic quantizers [3] are introduced to deal with the bandwidth constraints, and the corresponding fusion estimation error system model is established. By using the discrete-time bounded real lemma, a convex optimization problem on the choices of the optimal weighting matrices and quantization parameters is established in terms of linear matrix inequalities (LMIs). Moreover, it is proved that the performance of the designed fusion estimator is better than that of each local quantized estimator.

Distributed fusion is a typical structure for multisensor fusion estimation in WSNs, where the sensors generate local estimates ahead and then send them to a fusion center (FC) for fusion estimation [1, 2]. When the number of sensors is large, it is wasteful to embed in each sensor an estimator, and the FC requires a large bandwidth to communicate with the various sensors in a short time, which is usually impossible since the WSN is limited in bandwidth. An improvement is to adopt a hierarchical structure for fusion estimation [3–6]. In a hierarchical fusion estimation system, the sensors are divided into several clusters, and the sensors within the same cluster are connected to a local estimator. Moreover, only the local estimators are linked to the FC, and the measurements from sensors in a cluster are pretreated by local estimators in advance. A structure of the hierarchical fusion system is shown in Fig. 7.1. There are mainly two deficiencies in the existing hierarchical fusion estimation. First, local estimations and the fusion estimation are assumed to be time synchronized, which is restrictive as the processing rates of different clusters may be different from each other. Second, during the estimation interval, each sensor communicates with the local estimator only once, which implies that only one measurement from a sensor can be used for local estimation.