Search

In a networked multiaxis motion control task, faults in any motor will cause the performance degradation of cooperative operation, which may considerably affect the whole network and the quality of products. The main objective of this article is to propose an improved observer-based fault-tolerant tracking control approach for industrial multiagent systems. First, a group of new distributed intermediate estimators is presented, where the design structure is modified to enhance the feasibility of the estimation scheme. It is shown that both of the nominal distributed intermediate estimator and the traditional extended state observer are special cases of the proposed estimator. Second, the estimation performance can be improved significantly via an online reinforcement learning estimation strategy, whose core is an adaptive switching mechanism integrated with a function block of source fault mode localization. Benefiting from satisfactory estimation results, good fault-tolerant tracking control performance can be guaranteed despite of multiple faults and disturbances. The application to a networked multiaxis motion control system demonstrates the effectiveness and superiority of the proposed method. [ABSTRACT FROM AUTHOR]

Learning-from-crowds aims to design proper aggregation strategies to infer the unknown true labels from the noisy labels provided by ordinary web workers. This paper presents max-margin majority voting (M$^3$3V) to improve the discriminative ability of majority voting and further presents a Bayesian generalization to incorporate the flexibility of generative methods on modeling noisy observations with worker confusion matrices for different application settings. We first introduce the crowdsourcing margin of majority voting, then we formulate the joint learning as a regularized Bayesian inference (RegBayes) problem, where the posterior regularization is derived by maximizing the margin between the aggregated score of a potential true label and that of any alternative label. Our Bayesian model naturally covers the Dawid-Skene estimator and M$^3$3V as its two special cases. Due to the flexibility of our model, we extend it to handle crowdsourced labels with an ordinal structure with the main ideas about the crowdsourcing margin unchanged. Moreover, we consider an online learning-from-crowds setting where labels coming in a stream. Empirical results demonstrate that our methods are competitive, often achieving better results than state-of-the-art estimators. [ABSTRACT FROM AUTHOR]

Bearing remaining useful life (RUL) prediction plays a crucial role in guaranteeing safe operation of machinery and reducing maintenance loss. In this paper, we present a new deep feature learning method for RUL estimation approach through time frequency representation (TFR) and multiscale convolutional neural network (MSCNN). TFR can reveal nonstationary property of a bearing degradation signal effectively. After acquiring time-series degradation signals, we get TFRs, which contain plenty of useful information using wavelet transform. Owing to high dimensionality, the size of these TFRs is reduced by bilinear interpolation, which are further regarded as inputs for deep learning models. Here, we introduce an MSCNN model structure, which keeps the global and local information synchronously compared to a traditional convolutional neural network (CNN). The salient features, which contribute for RUL estimation, can be learned automatically by MSCNN. The effectiveness of the presented method is validated by the experiment data. Compared to traditional data-driven and different CNN-based feature extraction methods, the proposed method shows enhanced performance in the prediction accuracy. [ABSTRACT FROM AUTHOR]

Humans are avid consumers of visual content. Every day, people watch videos, play games, and share photos on social media. However, there is an asymmetry—while everybody is able to consume visual data, only a chosen few are talented enough to express themselves visually. For the rest of us, most attempts at creating realistic visual content end up quickly "falling off" what we could consider to be natural images. In this thesis, we investigate several machine learning approaches for preserving visual realism while creating and manipulating photographs. We use these methods as training wheels for visual content creation. These methods not only help users easily synthesize realistic photos but also enable previously not possible visual effects. [ABSTRACT FROM AUTHOR]

Deep generative models (DGMs) can effectively capture the underlying distributions of complex data by learning multilayered representations and performing inference. However, it is relatively insufficient to boost the discriminative ability of DGMs. This paper presents max-margin deep generative models (mmDGMs) and a class-conditional variant (mmDCGMs), which explore the strongly discriminative principle of max-margin learning to improve the predictive performance of DGMs in both supervised and semi-supervised learning, while retaining the generative capability. In semi-supervised learning, we use the predictions of a max-margin classifier as the missing labels instead of performing full posterior inference for efficiency; we also introduce additional max-margin and label-balance regularization terms of unlabeled data for effectiveness. We develop an efficient doubly stochastic subgradient algorithm for the piecewise linear objectives in different settings. Empirical results on various datasets demonstrate that: (1) max-margin learning can significantly improve the prediction performance of DGMs and meanwhile retain the generative ability; (2) in supervised learning, mmDGMs are competitive to the best fully discriminative networks when employing convolutional neural networks as the generative and recognition models; and (3) in semi-supervised learning, mmDCGMs can perform efficient inference and achieve state-of-the-art classification results on several benchmarks. [ABSTRACT FROM AUTHOR]

In this paper, we consider a heterogenous network in which low-power indoor femtocells are deployed in the coverage area of the existing macrobase station (MBS). This paper proposes a novel coordinated random beamforming and user-scheduling strategy to improve the throughput of users served by the femtocell access point (FAP) while satisfying the quality-of-service (QoS) requirements of users served by both MBS and FAP. The strategy, termed as QoS-aware coordinated scheduling (QACS), requires limited coordination between the MBS and FAP, i.e., only the indexes of the qualified beams are shared. Exact statistical analysis for the ergodic achievable rate of both FAP and MBS with the proposed strategy are presented. Scheduling fairness is also addressed for the proposed QACS. [ABSTRACT FROM PUBLISHER]

In future practical deployments of massive multi-input multi-output (MIMO) systems, the number of radio-frequency (RF) chains at the base stations (BSs) may be much smaller than the number of BS antennas to reduce the overall expenditure. In this correspondence, we propose a novel design framework for joint data and artificial noise (AN) precoding in a multiuser massive MIMO system with limited number of RF chains, which improves the wireless security performance. With imperfect channel state information (CSI), we analytically derive an achievable lower bound on the ergodic secrecy rate of any mobile terminal (MT) for both analog and hybrid precoding schemes. The closed-form lower bound is used to determine optimal power splitting between data and AN that maximizes the secrecy rate through simple 1-D search. Analytical and numerical results together reveal that the proposed hybrid precoder, although suffering from reduced secrecy rate compared with the theoretical full-dimensional precoder, is free of the high computational complexity of large-scale matrix inversion and null-space calculations and largely reduces the hardware cost. [ABSTRACT FROM AUTHOR]

The temperature coefficient of resonant frequency of a microwave dielectric material ( \tau f,s ) is defined as the \tau f of an ideal resonant system that is homogeneously filled with this material, while it is usually measured as the \tau f of a nonideal resonant system ( \tau f,r ) such as a cavity dielectric resonator. In the present work, finite element analysis is used to investigate the measurement error of \tau f,s by \mathrmTE01\delta -mode resonant cavity method, for which the sample is placed on a support and far from the cavity walls. Several error sources contribute to the total measurement error, which is usually several ppm/°C and decreases with the dielectric constant of the sample ( \varepsilon r,s ). Besides, the measurement error is strongly dependent on the cavity size. In comparison, the measurement error is much lower and usually within 1 ppm/°C for TE011-mode parallel plate method, where the sample is clamped between two metal plates. It is suggested to measure \tau f,s by TE011-mode parallel plate method but not by \mathrmTE01\delta -mode resonant cavity method if \tau f,s is treated as \tau f,r , especially when \varepsilon r,s is low and high measurement accuracy is needed. If \mathrmTE01\delta -mode resonant cavity method is used to measure \tau f,s , the result must be corrected with the aid of numerical calculation to improve the measurement accuracy. [ABSTRACT FROM PUBLISHER]

As advances in scientific and business data collection have exponentially created more data, electric utility companies are seeking new tools and techniques to turn the collected data into operational insights and assist with cost-saving decisions. The cost of building a specific business-driven big data application, however, can be tremendously high. This paper proposes developing a standard-based software framework to address key utility big data issues and foster development of big data analytical applications. Based on the support of this generic framework, new big data analytical solutions can be rapidly built and deployed to improve business practices in a utility organization. The proposed framework-based approach, as demonstrated in the conducted case studies, has proven to be promising for addressing the emerging big data challenges in the utility industry. [ABSTRACT FROM AUTHOR]

Relational topic models (RTMs) provide a probabilistic generative process to describe both the link structure and document contents for document networks, and they have shown promise on predicting network structures and discovering latent topic representations. However, existing RTMs have limitations in both the restricted model expressiveness and incapability of dealing with imbalanced network data. To expand the scope and improve the inference accuracy of RTMs, this paper presents three extensions: 1) unlike the common link likelihood with a diagonal weight matrix that allows the-same-topic interactions only, we generalize it to use a full weight matrix that captures all pairwise topic interactions and is applicable to asymmetric networks; 2) instead of doing standard Bayesian inference, we perform regularized Bayesian inference (RegBayes) with a regularization parameter to deal with the imbalanced link structure issue in real networks and improve the discriminative ability of learned latent representations; and 3) instead of doing variational approximation with strict mean-field assumptions, we present collapsed Gibbs sampling algorithms for the generalized relational topic models by exploring data augmentation without making restricting assumptions. Under the generic RegBayes framework, we carefully investigate two popular discriminative loss functions, namely, the logistic log-loss and the max-margin hinge loss. Experimental results on several real network datasets demonstrate the significance of these extensions on improving prediction performance. [ABSTRACT FROM AUTHOR]

In this paper, we tackle the problem of common object (multiple classes) discovery from a set of input images, where we assume the presence of one object class in each image. This problem is, loosely speaking, unsupervised since we do not know a priori about the object type, location, and scale in each image. We observe that the general task of object class discovery in a fully unsupervised manner is intrinsically ambiguous; here we adopt saliency detection to propose candidate image windows/patches to turn an unsupervised learning problem into a weakly-supervised learning problem. In the paper, we propose an algorithm for simultaneously localizing objects and discovering object classes via bottom-up (saliency-guided) multiple class learning (bMCL). Our contributions are three-fold: (1) we adopt saliency detection to convert unsupervised learning into multiple instance learning, formulated as bottom-up multiple class learning (bMCL); (2) we propose an integrated framework that simultaneously performs object localization, object class discovery, and object detector training; (3) we demonstrate that our framework yields significant improvements over existing methods for multi-class object discovery and possess evident advantages over competing methods in computer vision. In addition, although saliency detection has recently attracted much attention, its practical usage for high-level vision tasks has yet to be justified. Our method validates the usefulness of saliency detection to output “noisy input” for a top-down method to extract common patterns. [ABSTRACT FROM PUBLISHER]

We propose a unified game-theoretical framework to perform classification and conditional image generation given limited supervision. It is formulated as a three-player minimax game consisting of a generator, a classifier and a discriminator, and therefore is referred to as Triple Generative Adversarial Network (Triple-GAN). The generator and the classifier characterize the conditional distributions between images and labels to perform conditional generation and classification, respectively. The discriminator solely focuses on identifying fake image-label pairs. Theoretically, the three-player formulation guarantees consistency. Namely, under a nonparametric assumption, the unique equilibrium of the game is that the distributions characterized by the generator and the classifier converge to the data distribution. As a byproduct of the three-player formulation, Triple-GAN is flexible to incorporate different semi-supervised classifiers and GAN architectures. We evaluate Triple-GAN in two challenging settings, namely, semi-supervised learning and the extremely low data regime. In both settings, Triple-GAN can achieve excellent classification results and generate meaningful samples in a specific class simultaneously. In particular, using a commonly adopted 13-layer CNN classifier, Triple-GAN outperforms extensive semi-supervised learning methods substantially on several benchmarks no matter data augmentation is applied or not. [ABSTRACT FROM AUTHOR]

We propose a unified game-theoretical framework to perform classification and conditional image generation given limited supervision. It is formulated as a three-player minimax game consisting of a generator, a classifier and a discriminator, and therefore is referred to as Triple Generative Adversarial Network (Triple-GAN). The generator and the classifier characterize the conditional distributions between images and labels to perform conditional generation and classification, respectively. The discriminator solely focuses on identifying fake image-label pairs. Theoretically, the three-player formulation guarantees consistency. Namely, under a nonparametric assumption, the unique equilibrium of the game is that the distributions characterized by the generator and the classifier converge to the data distribution. As a byproduct of the three-player formulation, Triple-GAN is flexible to incorporate different semi-supervised classifiers and GAN architectures. We evaluate Triple-GAN in two challenging settings, namely, semi-supervised learning and the extremely low data regime. In both settings, Triple-GAN can achieve excellent classification results and generate meaningful samples in a specific class simultaneously. In particular, using a commonly adopted 13-layer CNN classifier, Triple-GAN outperforms extensive semi-supervised learning methods substantially on several benchmarks no matter data augmentation is applied or not. [ABSTRACT FROM AUTHOR]

Vacuum surface flashover of the high gradient insulator (HGI), which consists of many alternative layers of conductors and insulators, was found to generally start from one or several dielectric layers adjacent to the anode under nanosecond pulse by using the fast photography. A simple model is proposed to explain the experimental observations. The secondary electron emission avalanche (SEEA) along the dielectric surface adjacent to the anode will be greatly enhanced due to the upstream SEEA starting from the traditional and the intermediate cathode triple junctions. The 15-mm-thick HGI samples with different numbers of dielectric layers were tested and the samples with six and eight dielectric layers showed the best performance. The experimental result proves that as the number of dielectric layers reaches a certain level, the performance of the HGI will decrease as the number of dielectric layers increases. [ABSTRACT FROM PUBLISHER]

For images, gradient domain composition methods like Poisson blending offer practical solutions for uncertain object boundaries and differences in illumination conditions. However, adapting Poisson image blending to video presents new challenges due to the added temporal dimension. In video, the human eye is sensitive to small changes in blending boundaries across frames and slight differences in motions of the source patch and target video. We present a novel video blending approach that tackles these problems by merging the gradient of source and target videos and optimizing a consistent blending boundary based on a user-provided blending trimap for the source video. Our approach extends mean-value coordinates interpolation to support hybrid blending with a dynamic boundary while maintaining interactive performance. We also provide a user interface and source object positioning method that can efficiently deal with complex video sequences beyond the capabilities of alpha blending. [ABSTRACT FROM PUBLISHER]

Using shape information has gained increasing concerns in the task of image labeling. In this paper, we present a dynamic hybrid Markov random field (DHMRF), which explicitly captures middle-level object shape and low-level visual appearance (e.g., texture and color) for image labeling. Each node in DHMRF is described by either a deformable template or an appearance model as visual prototype. On the other hand, the edges encode two types of intersections: co-occurrence and spatial layered context, with respect to the labels and prototypes of connected nodes. To learn the DHMRF model, an iterative algorithm is designed to automatically select the most informative features and estimate model parameters. The algorithm achieves high computational efficiency since a branch-and-bound schema is introduced to estimate model parameters. Compared with previous methods, which usually employ implicit shape cues, our DHMRF model seamlessly integrates color, texture, and shape cues to inference labeling output, and thus produces more accurate and reliable results. Extensive experiments validate its superiority over other state-of-the-art methods in terms of recognition accuracy and implementation efficiency on: 1) the MSRC 21-class dataset, and 2) the lotus hill institute 15-class dataset. [ABSTRACT FROM AUTHOR]

We study straight-crested waves and vibration modes with spatial variations along the x3 direction only in an AT-cut quartz plate resonator. The equations of anisotropic elasticity are used. Dispersion relations for face-shear and thickness-twist waves in unbounded plates are plotted. Frequency spectra are obtained for face-shear and thickness-twist vibrations of finite plates in which these modes are coupled by boundary conditions. Most importantly, our analysis produces the frequency spectra for overtone modes which do not seem to have been obtained before for x3-dependent modes. Numerical results for third- and fifth-overtone AT-cut quartz resonators are presented, showing that higher-order overtone modes are associated with more mode couplings. [ABSTRACT FROM AUTHOR]

Learning salient representations of multiview data is an essential step in many applications such as image classification, retrieval, and annotation. Standard predictive methods, such as support vector machines, often directly use all the features available without taking into consideration the presence of distinct views and the resultant view dependencies, coherence, and complementarity that offer key insights to the semantics of the data, and are therefore offering weak performance and are incapable of supporting view-level analysis. This paper presents a statistical method to learn a predictive subspace representation underlying multiple views, leveraging both multiview dependencies and availability of supervising side-information. Our approach is based on a multiview latent subspace Markov network (MN) which fulfills a weak conditional independence assumption that multiview observations and response variables are conditionally independent given a set of latent variables. To learn the latent subspace MN, we develop a large-margin approach which jointly maximizes data likelihood and minimizes a prediction loss on training data. Learning and inference are efficiently done with a contrastive divergence method. Finally, we extensively evaluate the large-margin latent MN on real image and hotel review datasets for classification, regression, image annotation, and retrieval. Our results demonstrate that the large-margin approach can achieve significant improvements in terms of prediction performance and discovering predictive latent subspace representations. [ABSTRACT FROM PUBLISHER]

Information visualization appears to be a promising technique for improving the business practices in today's electric power industry. The legacy power system visualization tools, however, restrict the visualization process to follow a limited number of pre-defined patterns created by human designers, thus hindering users' ability to discover. This paper proposes a data-driven approach to interactive visualization of power systems. The proposed approach relies on developing powerful data manipulation algorithms to create visualizations based on the characteristics of empirically or mathematically derived data. Based on this approach, a data-driven model exploratory tool has been developed to enable users to visualize the power system's physical/electrical configurations at various levels and from different perspectives. The conducted case studies have demonstrated that the data-driven approach could result in an interactive and user-driven power system visualization tool that fosters scientific understanding and insight, therefore unleashing the power of visualization. [ABSTRACT FROM PUBLISHER]

The typical response of an X-ray converter material to the passage of a high-current relativistic electron beam is ionization and rapid dispersal. Experiments have been performed on the Dragon-I linear induction accelerator to study this phenomenon. In this paper, we report the fast photography and Faraday cup experiments performed to quantify the plasma-plume expansion and velocities of ions. Monte Carlo simulation shows that the energy deposition in a multilayer target is about 7 kJ/g. By solving the Saha equation, the ionization degree of the plasma was figured out to be only 14%. [ABSTRACT FROM PUBLISHER]

Provides information on a study which introduced the concepts of design patterns in object-oriented modeling and applying this technique for power system modeling. Overview of design patterns; Problems on power system modeling and how to apply design patterns; Conclusions of the study.

Explores the application of object-oriented technology throughout the development of the software systems for power system simulations. Overview of the object modeling technique; Development of a distribution analysis system using the object modeling technique; Conclusions.

Presents techniques for locating and diagnosing faults on electric power distribution feeders. Identification of the actual fault location; Distribution system protection; Phase current and load changes on the faulted phase.

Proposes a digital sampling technique for measuring amplitude and playback waveforms relative to full bandwidth sampling in the magnetic head testing. Implementation of the digital sampling technique; Simulation results; Experimental results.

Presents information on a study of the amplitude and asymmetry correlation between quasi-static and dynamic electrical testing of magnetoresistive sensors. Methodology used in the study; Results of the study.

Studies the modeling of lamination effects for thin-film inductive heads. Comparison of impedance results with experimental data; Isolated pulse for the laminated heads; Modeled head efficiency.

Characteristic response parameterized in time-dependent target intrinsic polarizabilities has been widely used to create feature libraries for unexploded ordnance (UXO) discrimination. This paper discusses the accurate measurement of characteristic response for UXOs with transient electromagnetic (TEM) system. First, a specially designed TEM system, which utilizes a modified solenoid and a pair of rectangular coils as transmitting coils to ensure the uniformity of primary field up to 96% throughout the target, is described. Then, the characteristic responses of 12 UXOs are measured from directions vertical or parallel to the axis at distances from 40 to 120 cm. The measured characteristic responses show a power law decay function at an early time and an exponential decay function at a later time. The characteristic responses parallel to the target axis are larger, and the decay is considerably slower than those vertical to the axis, leading to one to two orders of magnitude difference at a later time. Results also show that the characteristic response measured from the direction vertical to the axis of the target is more precise than that from the direction parallel to the axis at the same distance. The portion of the target that is close to the system accounts for a large proportion of the response. A target can be approximated as a single dipole only when it is located no less than twice its length from the system. Finally, the characteristic responses of 12 targets are parameterized with the Pasion–Oldenburg expression for feature library creation. [ABSTRACT FROM AUTHOR]

In this paper, we investigate secure transmission in a massive multiple-input multiple-output system adopting low-resolution digital-to-analog converters (DACs). Artificial noise (AN) is deliberately transmitted simultaneously with the confidential signals to degrade the eavesdropper’s channel quality. By applying the Bussgang theorem, a DAC quantization model is developed which facilitates the analysis of the asymptotic achievable secrecy rate. Interestingly, for a fixed power allocation factor $\phi $ , low-resolution DACs typically result in a secrecy rate loss, but in certain cases, they provide superior performance, e.g., at low signal-to-noise ratio (SNR). Specifically, we derive a closed-form SNR threshold which determines whether low-resolution or high-resolution DACs are preferable for improving the secrecy rate. Furthermore, a closed-form expression for the optimal $\phi $ is derived. With AN generated in the null-space of the user channel and the optimal $\phi $ , low-resolution DACs inevitably cause secrecy rate loss. On the other hand, for random AN with the optimal $\phi $ , the secrecy rate is hardly affected by the DAC resolution because the negative impact of the quantization noise can be compensated by reducing the AN power. All the derived analytical results are verified by numerical simulations. [ABSTRACT FROM AUTHOR]

In this paper, we study the resource allocation algorithm design for multiple-input single-output (MISO) multicarrier non-orthogonal multiple access (MC-NOMA) systems, in which a full-duplex base station serves multiple half-duplex uplink and downlink users on the same subcarrier simultaneously. The resource allocation is optimized for maximization of the weighted system throughput while the information leakage is constrained and artificial noise is injected to guarantee secure communication in the presence of multiple potential eavesdroppers. To this end, we formulate a robust non-convex optimization problem taking into account the imperfect channel state information of the eavesdropping channels and the quality-of-service requirements of the legitimate users. Despite the non-convexity of the optimization problem, we solve it optimally by applying monotonic optimization which yields the optimal beamforming, artificial noise design, subcarrier allocation, and power allocation policy. The optimal resource allocation policy serves as a performance benchmark since the corresponding monotonic optimization-based algorithm entails a high computational complexity. Hence, we also develop a low-complexity suboptimal resource allocation algorithm which converges to a locally optimal solution. Our simulation results reveal that the performance of the suboptimal algorithm closely approaches that of the optimal algorithm. Besides, the proposed optimal MISO NOMA system can not only ensure downlink and uplink communication security simultaneously but also provides a significant system secrecy rate improvement compared with the traditional MISO orthogonal multiple access systems and two other baseline schemes. [ABSTRACT FROM AUTHOR]

Person reidentification matches person observations captured by nonoverlapping cameras at different times and locations. Due to the appearance variations caused by view angle, pose, lighting, background, and occlusion, person reidentification is a challenging task. To improve the accuracy of person reidentification, this paper proposes a metric learning method. The main idea of our method is to learn a low-dimensional metric space, in which the features extracted from different observations of the same person are pulled, and the features extracted from observations of different persons but in the neighborhood of original feature space are pushed. Then all the features are projected to this metric space for matching. Compared with other metric learning methods, our method differentiates negative samples according to their importances in constructing the metric space, and pays more attention to restrain the more confusable negative pairs. Moreover, our method can simultaneously learn a uniform metric space for multiple features that is practical for multifeature fusion, an effective way to improve reidentification performance. Experimental results on three popular data sets show that our method is more robust against appearance changes and it outperforms other state-of-the-art reidentification methods. [ABSTRACT FROM AUTHOR]

Supervised topic models simultaneously model the latent topic structure of large collections of documents and a response variable associated with each document. Existing inference methods are based on variational approximation or Monte Carlo sampling, which often suffers from the local minimum defect. Spectral methods have been applied to learn unsupervised topic models, such as latent Dirichlet allocation (LDA), with provable guarantees. This paper investigates the possibility of applying spectral methods to recover the parameters of supervised LDA (sLDA). We first present a two-stage spectral method, which recovers the parameters of LDA followed by a power update method to recover the regression model parameters. Then, we further present a single-phase spectral algorithm to jointly recover the topic distribution matrix as well as the regression weights. Our spectral algorithms are provably correct and computationally efficient. We prove a sample complexity bound for each algorithm and subsequently derive a sufficient condition for the identifiability of sLDA. Thorough experiments on synthetic and real-world datasets verify the theory and demonstrate the practical effectiveness of the spectral algorithms. In fact, our results on a large-scale review rating dataset demonstrate that our single-phase spectral algorithm alone gets comparable or even better performance than state-of-the-art methods, while previous work on spectral methods has rarely reported such promising performance. [ABSTRACT FROM AUTHOR]

Adversarial attacks have been extensively studied in recent years since they can identify the vulnerability of deep learning models before deployed. In this paper, we consider the black-box adversarial setting, where the adversary needs to craft adversarial examples without access to the gradients of a target model. Previous methods attempted to approximate the true gradient either by using the transfer gradient of a surrogate white-box model or based on the feedback of model queries. However, the existing methods inevitably suffer from low attack success rates or poor query efficiency since it is difficult to estimate the gradient in a high-dimensional input space with limited information. To address these problems and improve black-box attacks, we propose two prior-guided random gradient-free (PRGF) algorithms based on biased sampling and gradient averaging, respectively. Our methods can take the advantage of a transfer-based prior given by the gradient of a surrogate model and the query information simultaneously. Through theoretical analyses, the transfer-based prior is appropriately integrated with model queries by an optimal coefficient in each method. Extensive experiments demonstrate that, in comparison with the alternative state-of-the-arts, both of our methods require much fewer queries to attack black-box models with higher success rates. [ABSTRACT FROM AUTHOR]

Conditional Generative Adversarial Networks (cGANs) have enabled controllable image synthesis for many vision and graphics applications. However, recent cGANs are 1-2 orders of magnitude more compute-intensive than modern recognition CNNs. For example, GauGAN consumes 281G MACs per image, compared to 0.44G MACs for MobileNet-v3, making it difficult for interactive deployment. In this work, we propose a general-purpose compression framework for reducing the inference time and model size of the generator in cGANs. Directly applying existing compression methods yields poor performance due to the difficulty of GAN training and the differences in generator architectures. We address these challenges in two ways. First, to stabilize GAN training, we transfer knowledge of multiple intermediate representations of the original model to its compressed model and unify unpaired and paired learning. Second, instead of reusing existing CNN designs, our method finds efficient architectures via neural architecture search. To accelerate the search process, we decouple the model training and search via weight sharing. Experiments demonstrate the effectiveness of our method across different supervision settings, network architectures, and learning methods. Without losing image quality, we reduce the computation of CycleGAN by 21×, Pix2pix by 12×, MUNIT by 29×, and GauGAN by 9×, paving the way for interactive image synthesis. [ABSTRACT FROM AUTHOR]

Adversarial attacks have been extensively studied in recent years since they can identify the vulnerability of deep learning models before deployed. In this paper, we consider the black-box adversarial setting, where the adversary needs to craft adversarial examples without access to the gradients of a target model. Previous methods attempted to approximate the true gradient either by using the transfer gradient of a surrogate white-box model or based on the feedback of model queries. However, the existing methods inevitably suffer from low attack success rates or poor query efficiency since it is difficult to estimate the gradient in a high-dimensional input space with limited information. To address these problems and improve black-box attacks, we propose two prior-guided random gradient-free (PRGF) algorithms based on biased sampling and gradient averaging, respectively. Our methods can take the advantage of a transfer-based prior given by the gradient of a surrogate model and the query information simultaneously. Through theoretical analyses, the transfer-based prior is appropriately integrated with model queries by an optimal coefficient in each method. Extensive experiments demonstrate that, in comparison with the alternative state-of-the-arts, both of our methods require much fewer queries to attack black-box models with higher success rates. [ABSTRACT FROM AUTHOR]

Conditional Generative Adversarial Networks (cGANs) have enabled controllable image synthesis for many vision and graphics applications. However, recent cGANs are 1-2 orders of magnitude more compute-intensive than modern recognition CNNs. For example, GauGAN consumes 281G MACs per image, compared to 0.44G MACs for MobileNet-v3, making it difficult for interactive deployment. In this work, we propose a general-purpose compression framework for reducing the inference time and model size of the generator in cGANs. Directly applying existing compression methods yields poor performance due to the difficulty of GAN training and the differences in generator architectures. We address these challenges in two ways. First, to stabilize GAN training, we transfer knowledge of multiple intermediate representations of the original model to its compressed model and unify unpaired and paired learning. Second, instead of reusing existing CNN designs, our method finds efficient architectures via neural architecture search. To accelerate the search process, we decouple the model training and search via weight sharing. Experiments demonstrate the effectiveness of our method across different supervision settings, network architectures, and learning methods. Without losing image quality, we reduce the computation of CycleGAN by 21×, Pix2pix by 12×, MUNIT by 29×, and GauGAN by 9×, paving the way for interactive image synthesis. [ABSTRACT FROM AUTHOR]

Damaged building detection from remote sensing imagery helps to quickly and rapidly assess losses after an earthquake. In recent years, deep learning technology has become a favorable tool for remote sensing image information detection. Based on the characteristics of damaged buildings in remote sensing images, in this article, a framework for damaged building detection that considers heterogeneity characteristics is proposed. First, a local–global context attention module is proposed to improve the feature detection ability of the network, which can extract the features of damaged buildings from different directions and effectively aggregate global and local features. In addition, the module takes the correlation between feature maps at different scales into account while extracting information. Second, a feature fusion module with self-attention is established to replace the simple connection between the encoding and decoding processes, which improves the detail feature recovery ability of the network during the upsampling process. Finally, to fully aggregate semantic and detailed features at different scales, a multibranch auxiliary classifier is established by adding two separate branches in the prediction stage. The effectiveness of the proposed approach is verified based on the data from the 2010 Haiti earthquake, and comparisons with three object-oriented methods and 16 existing excellent deep learning models are performed. The intersection over union (IOU) increase of 0.03%–7.39% is achieved using the proposed approach compared with excellent deep learning models. [ABSTRACT FROM AUTHOR]

Opportunistic scheduling schemes are investigated for uplink wiretap channels with multiple asymmetrically located legitimate users (LUs) and eavesdroppers. To exploit multiuser diversity, the cumulative distribution function-based scheduling method is leveraged to schedule the transmissions of the LUs. Under this scheduling framework, the closed-form expressions of the secrecy outage probability and ergodic secrecy rate are derived, illustrating the interplay among the system parameters, such as the channel statistics and the number of LUs and eavesdroppers. Through the secrecy outage analysis of the proposed scheduling schemes, we observe that the secrecy throughput is not always maximized with a larger channel access ratio (CAR), and consequently, we design a CAR adjustment scheme to maximize the secrecy throughput while satisfying the required secrecy level. We also prove that under our proposed scheduling schemes, the secrecy diversity order of each LU is equal to the reciprocal of the LU's CAR, implying that full diversity order is achieved, and the ergodic secrecy rate of each LU normalized by its CAR achieves the optimal double-logarithmic growth when the number of LUs increases to infinity. [ABSTRACT FROM AUTHOR]

As the successors of the overdue HuanjingJianzai-1 (HJ-1) satellites and new members in Chinese Environmental Protection and Disaster Monitoring Satellite Constellation, the first two of HuanjingJianZai-2 series satellites (HJ-2 A/B) have been launched on September 27, 2020. Each satellite carries four sensors, including the polarized scanning atmospheric corrector (PSAC), the charge-coupled device (CCD) camera, the hyperspectral imager (HSI), and the infrared spectroradiometer (IRS). Among them, PSAC is mainly used for the monitoring of atmospheric parameters to provide data support for atmospheric environmental monitoring and atmospheric correction of data from other sensors. To test the in-orbit performance of PSAC, we develop the “day-1” aerosol and water vapor retrieval algorithms. The preliminary validation results based on ground-based observations show that the aerosol optical depth (AOD) and columnar water vapor (CWV) datasets developed based on PSAC data have high accuracy and can effectively characterize the temporal trends of AOD and CWV. The accuracy of PSAC AOD dataset is better than the expected error (EE) ±(0.05 + 0.2 * AODAERONET), and the accuracy of PSAC CWV dataset is better than the EE ±(0.05 + 0.15 * CWVAERONET). To eliminate the negative impact of the atmosphere on CCD data and expand its application range, aerosol and water vapor data developed based on PSAC are used for atmospheric correction of CCD data. Compared with L1 CCD data, the texture details and clarity of CCD data after atmospheric correction have been significantly improved. [ABSTRACT FROM AUTHOR]

The diversity of magnetic field information obtained in marine geomagnetic measurements is of great relevance to defining detailed features of the targets. In this article, a solution algorithm based on the combination of scalar and vector fields is proposed. The magnetic direction of the geomagnetic field was obtained by repeatedly applying bias fields, and the multiparameter integrated measurement is realized based on the geometric relationship of the geomagnetic parameters. To make an instrument feasible for undersea measurement, a miniaturized multiparameter magnetometer was constructed within the limited space of a pressure tank, using an Overhauser scalar sensor and a spherical coil for generating bias fields. Test platforms were built to evaluate the scalar sensor and spherical coil. Results from a nonmagnetic laboratory showed that the expanded uncertainty of the scalar sensor was 0.11 nT, whereas the magnetic field generated by the spherical coil had 97% uniformity and completely engulfed the scalar sensor. Moreover, field experiments indicated that the standard deviation of prototype instrument in magnetic direction (3 $^{\prime \prime }$) was close to that of declination–inclination magnetometer, whereas its standard deviation in Z-component of 0.24 nT was better than that of a commercial fluxgate magnetometer. It also had markedly better measurement stability, with no temperature drift. [ABSTRACT FROM AUTHOR]

We study the propagation of thickness-twist and face-shear waves in piezoelectric plates of monoclinic crystals, which include rotated Y-cut quartz and langasite as special cases. The equations of linear piezoelectricity are used. Exact solutions are obtained for both fully-electroded and unelectroded plates. Dispersion relations are plotted for both AT-cut quartz and Y-cut langasite plates. The difference in frequencies between the exact piezoelectric solutions obtained in the present paper and the approximate elastic solutions in the literature is found to be of the order of 1%, which is significant in acoustic wave resonator and sensor applications. [ABSTRACT FROM PUBLISHER]

Hypervisor-based fault tolerance (HBFT), which synchronizes the state between the primary VM and the backup VM at a high frequency of tens to hundreds of milliseconds, is an emerging approach to sustaining mission-critical applications. Based on virtualization technology, HBFT provides an economic and transparent fault tolerant solution. However, the advantages currently come at the cost of substantial performance overhead during failure-free, especially for memory intensive applications. This paper presents an in-depth examination of HBFT and options to improve its performance. Based on the behavior of memory accesses among checkpointing epochs, we introduce two optimizations, read-fault reduction and write-fault prediction, for the memory tracking mechanism. These two optimizations improve the performance by 31 percent and 21 percent, respectively, for some applications. Then, we present software superpage which efficiently maps large memory regions between virtual machines (VM). Our optimization improves the performance of HBFT by a factor of 1.4 to 2.2 and achieves about 60 percent of that of the native VM. [ABSTRACT FROM AUTHOR]

Typical perovskite oxides SrTiO3 (STO) and PbZr0.52Ti0.48O3 (PZT) were fabricated on GaN semiconductor substrates by pulsed laser deposition. STO and PZT films were deposited on bare GaN, TiO2, and MgO-buffered GaN. The effects of TiO2 and MgO buffer-layers on the orientations and electric properties of the perovskite films were systematically studied. The crystalline properties were in situ monitored by reflective high energy electron diffraction and ex situ characterized by X-ray diffraction. It was found that the epitaxial temperature of STO and PZT was reduced by inserting a buffer layer. (111)-oriented films were obtained on bare and TiO2-buffered GaN. However, the orientations of the perovskite films were changed to be (110) when deposited on MgO buffer layer. Furthermore, PZT films deposited on MgO- and TiO2-buffered GaN show better electric performance compared with these films directly deposited on GaN. These results show that perovskite oxide could be epitaxially grown on GaN semiconductor substrates by inserting a proper buffer layer. [ABSTRACT FROM PUBLISHER]

This paper proposes a new method to select single-line-to-ground (SLG) fault feeder in resonant grounding power distribution system based on compensation characteristic of extinction coil in different frequencies. The proposed method is insusceptible to the influence of device polarity reverse installation. Firstly, through analyzing the zero sequence equivalent circuit under a SLG fault, the difference of zero sequence current share in fundamental frequency and second-harmonic frequency is investigated. And then, according to the change of current share (CCS), fault feeder selection criterion and steps are established. Finally, numerous simulation tests demonstrate the effectiveness of the proposed method in identifying the faulty feeder or the faulty bus. Besides, its reliability is also verified by tests with different system parameters and different fault conditions. Specially, device polarity reverse installation tests and comparisons are also implemented, which shows the advantages of the proposed method in dealing with device polarity reverse installation. [ABSTRACT FROM AUTHOR]

We are developing, at the Huazhong University of Science and Technology, a small-animal positron emission tomograph (PET) scanner that has two new features. Firstly, it uses detector modules that are based on a new digital readout electronics previously investigated and reported by us. Secondly, it can arrange the detector modules into various scanner geometries for different imaging needs. In this paper, we will describe the design and implementation of the detector module and report its performance properties. The detector module has a total detection sensitive area of 5.3\,\times\,5.3 cm^2, an overall energy resolution of 14.1% FWHM at 511 keV, and a block-level coincidence timing resolution of 1.5 ns FWHM. In addition, our preliminary imaging study shows that our small-animal PET scanner can resolve 1.0 mm-diameter rods separated by 2.0 mm. [ABSTRACT FROM PUBLISHER]

Partitioning nonlinearly separable datasets is a basic problem that is associated with data clustering. In this paper, a novel approach that is termed graph-based multiprototype competitive learning (GMPCL) is proposed to handle this problem. A graph-based method is employed to produce an initial, coarse clustering. After that, a multiprototype competitive learning is introduced to refine the coarse clustering and discover clusters of an arbitrary shape. The GMPCL algorithm is further extended to deal with high-dimensional data clustering, i.e., the fast graph-based multiprototype competitive learning (FGMPCL) algorithm. An experimental comparison has been performed by the exploitation of both synthetic and real-world datasets to validate the effectiveness of the proposed methods. Additionally, we apply our GMPCL/FGMPCL to two computer-vision tasks, namely, automatic color image segmentation and video clustering. Experimental results show that GMPCL/FGMPCL provide an effective and efficient tool with application to computer vision. [ABSTRACT FROM PUBLISHER]