Search

Multistatic SAR is able to observe targets from different angles simultaneously, which enhances the information acquiring capability. However, multistatic SAR can still be affected by electromagnetic jamming, resulting in the misinterpretation of multistatic SAR images. This article proposes a method to locate multiple range-deception jammers and suppress jamming signals. First, the echo model of multistatic SAR under a multijammer environment is established. Second, the detection of interested targets in multistatic SAR images can be achieved through visual saliency detection methods based on spectral residual. Third, location distribution features of false targets in multistatic SAR images are analyzed, and the Euclidean distance criteria are used to effectively distinguish false targets. Accurate localization is then achieved by combing multistatic SAR configuration information. Finally, using a linear constrained minimum variance beamforming algorithm to suppress jamming signals, multistatic SAR images without jamming signals can be obtained. Simulation results validate the effectiveness of the proposed method in this article.

The prevalence of online user-generated content has attracted great interest in textual sentiment analysis, which provides a low-cost yet effective way to discern consumers and markets. A mainstream of sentiment analysis is to construct a classification model with Bag-of-Words (BoW) features, but the large vocabulary base and skewed distribution of term frequency consistently pose research challenges, which is made even worse by the limited valid sentiment labels. In light of this, in this paper, we propose a novel method called Structural Holes based Sentiment Classifier (SHSC) for BoW-based sentiment classification. The key to SHSC is to reinforce the classification contribution of semantically rich words with clear-cut sentiment polarity. To this end, a word co-occurrence network is carefully constructed to represent both high and low frequency words. The work to find classification-inefficient words is then transformed into the identification of so-called bridge nodes that occupy the positions of structural holes in the network. Two interesting measures, i.e., information advantage rank and control advantage weight, are then designed elaborately for this purpose, which are based on the proposed sentiment-label propagation and short-path computation algorithms, respectively. SHSC finally feeds this information as the key regularizers into a simple regression model to guide parametric learning. Extensive experiments on real-world text datasets demonstrate the advantage of our SHSC model over competitive benchmarks, particularly when sentiment labels are scarce. The effectiveness of uncovering structural holes for sentiment classification is also carefully verified with some robustness checks and demonstration cases.

The fuzzy cognitive map (FCM) is a powerful model for system state prediction and interpretable knowledge representation. Recent years have witnessed the tremendous efforts devoted to enhancing the basic FCM, such as introducing temporal factors, uncertainty or fuzzy rules to improve interpretation, and introducing fuzzy neural networks or wavelets to improve time series prediction. But how to achieve high-precision yet interpretable prediction in cross-domain real-life applications remains a great challenge. In this article, we propose a novel FCM extension called deep FCM (DFCM) for multivariate time series forecasting, in order to take both the advantage of FCM in interpretation and the advantage of deep neural networks in prediction. Specifically, to improve the predictive power, DFCM leverages a fully connected neural network to model connections (relationships) among concepts in a system, and a recurrent neural network to model unknown exogenous factors that have influences on system dynamics. Moreover, to foster model interpretability encumbered by the embedded deep structures, a partial derivative-based approach is proposed to measure the connection strengths between concepts in DFCM. An alternate function gradient descent algorithm is then proposed for parameter inference. The effectiveness of DFCM is validated over four publicly available datasets with the presence of seven baselines. DFCM indeed provides an important clue to building interpretable predictors for real-life applications.

In synthetic aperture radar (SAR) imaging, moving target is generally mixed with stationary targets. Meanwhile, the image of a moving target is distorted and displaced due to the lack of its prior velocity information. Furthermore, imaging of a moving target for geosynchronous (GEO) spaceborne-airborne bistatic SAR (GEO SA-BiSAR) is a more challenging problem because the echo is sub-Nyquist sampled in azimuth. In this article, a simultaneous moving and stationary target imaging method for GEO SA-BiSAR is proposed. First, range models and the corresponding echo models of moving and stationary targets are established. The observation models for both moving and stationary targets with two receiving channels are derived based on the inverse of an efficient imaging algorithm. After that, the imaging problem of moving and stationary targets is modeled as a joint velocity estimation and sparse decomposition problem, which aims at optimizing the entropy of the moving target image and residual error of the formed images at the same time. Finally, a joint optimization method based on the particle swarm optimization (PSO) method and alternating direction method of multipliers (ADMM) is applied to achieve the imaging of moving and stationary targets and estimation of the moving target velocity. With two receiving channels, the accurate separation and focusing of stationary and moving targets as well as the precise estimation of moving target velocity can be achieved with sub-Nyquist sampling echo. Simulation results are presented to validate the effectiveness of the proposed method.

Traditional computation is driven by parallel accelerators or distributed computation nodes in order to improve computing performance, save energy, and decrease delays in accessing memory. Recently, artificial intelligent algorithms, frameworks, and computing models are growing to help with high computational performance.

Due to its ability of forming continuous images for a ground scene of interest, the video synthetic aperture radar (SAR) has been studied in recent years. However, as video SAR needs to reconstruct many frames, the data are of enormous amount and the imaging process is of large computational cost, which limits its applications. In this article, we exploit the redundancy property of multiframe video SAR data, which can be modeled as low-rank tensor, and formulate the video SAR imaging process as a low-rank tensor recovery problem, which is solved by an efficient alternating minimization method. We empirically compare the proposed method with several state-of-the-art video SAR imaging algorithms, including the fast back-projection (FBP) method and the compressed sensing (CS)-based method. Experiments on both simulated and real data show that the proposed low-rank tensor-based method requires significantly less amount of data samples while achieving similar or better imaging performance.

Whether investor sentiment affects stock prices is an issue of long-standing interest for economists. We conduct a comprehensive study of the predictability of investor sentiment, which is measured directly by extracting expectations from online user-generated content (UGC) on the stock message board of Eastmoney.com in the Chinese stock market. We consider the influential factors in prediction, including the selections of different text classification algorithms, price forecasting models, time horizons, and information update schemes. Using comparisons of the long short-term memory (LSTM) model, logistic regression, support vector machine, and Naive Bayes model, the results show that daily investor sentiment contains predictive information only for open prices, while the hourly sentiment has two hours of leading predictability for closing prices. Investors do update their expectations during trading hours. Moreover, our results reveal that advanced models, such as LSTM, can provide more predictive power with investor sentiment only if the inputs of a model contain predictive information.

In recent years, short text topic modeling has drawn considerable attentions from interdisciplinary researchers. Various customized topic models have been proposed to tackle the semantic sparseness nature of short texts. Most (if not all) of them follow the bag-of-words assumption, which, however, is not adequate since word order and phrases are often critical to capturing the meaning of texts. On the other hand, while some existing topic models are sensitive to word order, they do not perform well on short texts due to the severe data sparseness. To address these issues, we propose the Pseudo-document-based Topical N-Grams model (PTNG), which alleviates the data sparsity problem of short texts while is sensitive to word order. Extensive experiments on three real-world data sets with state-of-the-art baselines demonstrate the high quality of topics learned by PTNG according to UCI coherence scores and more discriminative semantic representation of short texts according to classification results.

It is always deemed crucial to identify the key factors that could have significant impact on the stock market trend. Recently, an interesting phenomenon has emerged that some of President Trump’s posts in Twitter can surge into a dominant role on the stock market for a certain time period, although studies along this line are still in their infancy. Therefore, in this article, we study whether and how this new-rising information can help boost the performance of stock market prediction. Specifically, we have found that the echoing reinforced effect of financial news with Trump’s market-related tweets can influence the market movement—that is, some of Trump’s tweets directly impact the stock market in a short time, and the impact can be further intensified when it echoes with other financial news reports. Along this line, we propose a deep information echoing model to predict the hourly stock market trend, such as the rise and fall of the Dow Jones Industrial Average. In particular, to model the discovered echoing reinforced impact, we design a novel information echoing module with a gating mechanism in a sequential deep learning framework to capture the fused knowledge from both Trump’s tweets and financial news. Extensive experiments have been conducted on the real-world U.S. stock market data to validate the effectiveness of our model and its interpretability in understanding the usability of Trump’s posts. Our proposed deep echoing model outperforms other baselines by achieving the best accuracy of 60.42% and obtains remarkable accumulated profits in a trading simulation, which confirms our assumption that Trump’s tweets contain indicative information for short-term market trends. Furthermore, we find that Trump’s tweets about trade and political events are more likely to be associated with short-term market movement, and it seems interesting that the impact would not degrade as time passes.

Lifelong topic model (LTM), an emerging paradigm for never-ending topic learning, aims to yield higher-quality topics as time passes through knowledge accumulated from the past yet learned for the future. In this paper, we propose a novel lifelong topic model based on non-negative matrix factorization (NMF), called Affinity Regularized NMF for LTM (NMF-LTM), which to our best knowledge is distinctive from the popular LDA-based LTMs. NMF-LTM achieves lifelong learning by introducing word-word graph Laplacian as semantic affinity regularization. Other priors such as sparsity, diversity, and between-class affinity are incorporated as well for better performance, and a theoretical guarantee is provided for the algorithmic convergence to a local minimum. Extensive experiments on various public corpora demonstrate the effectiveness of NMF-LTM, particularly its human-like behaviors in two carefully designed learning tasks and the ability in topic modeling of big data. A further exploration of semantic relatedness in knowledge graphs and a case study on a large-scale real-world corpus exhibit the strength of NMF-LTM in discovering high-quality topics in an efficient and robust way.

The correlation network, typically formed by computing pairwise correlations between variables, has recently become a competitive paradigm to discover insights in various application domains, such as climate prediction, financial marketing, and bioinformatics. In this study, we adopt this paradigm to detect cheating behavior hidden in business distribution channels, where falsified big deals are often made by collusive partners to obtain lower product prices—a behavior deemed to be extremely harmful to the sale ecosystem. To this end, we assume that abnormal deals are likely to occur between two partners if their purchase-volume sequences have a strong negative correlation. This seemingly intuitive rule, however, imposes several research challenges. First, existing correlation measures are usually symmetric and thus cannot distinguish the different roles of partners in cheating. Second, the tick-to-tick correspondence between two sequences might be violated due to the possible delay of purchase behavior, which should also be captured by correlation measures. Finally, the fact that any pair of sequences could be correlated may result in a number of false-positive cheating pairs, which need to be corrected in a systematic manner. To address these issues, we propose a correlation network analysis framework for cheating detection. In the framework, we adopt an asymmetric correlation measure to distinguish the two roles, namely, cheating seller and cheating buyer , in a cheating alliance. Dynamic Time Warping is employed to address the time offset between two sequences in computing the correlation. We further propose two graph-cut methods to convert the correlation network into a bipartite graph to rank cheating partners, which simultaneously helps to remove false-positive correlation pairs. Based on a 4-year real-world channel dataset from a worldwide IT company, we demonstrate the effectiveness of the proposed method in comparison to competitive baseline methods.

Fraud detection is usually compared to finding a needle in a haystack and remains a challenging task because fraudulent acts are buried in massive amounts of normal behavior and true intentions may be disguised in a single snapshot. Indeed, fraudulent incidents usually take place in consecutive time steps to gain illegal benefits, which provides unique clues for probing fraudulent behavior by considering a complete behavioral sequence rather than detecting fraud from a snapshot of behavior. Additionally, fraudulent behavior may involve different parties, such that the interaction patterns between sources and targets can help distinguish fraudulent acts from normal behavior. Therefore, in this paper, we model the attributed behavioral sequences generated from consecutive behaviors in order to capture the sequential patterns, while those that deviate from the pattern can be detected as fraudulence. Considering the characteristics of the behavioral sequence, we propose a novel model, NHA-LSTM, by augmenting the traditional LSTM with a modified forget gate, where the interval time between consecutive time steps is considered. Furthermore, we design a self-historical attention mechanism to allow for long time dependencies, which can help identify repeated or cyclical appearances. In addition, we propose an enhanced network embedding method, FraudWalk, to construct embeddings for the nodes in the interaction network with regard to higher-order interactions and particular time constraints for revealing potential group fraudulence. The node embeddings, along with the feature vectors, are fed into the model to capture the interactions between sources and targets. To validate the effectiveness of sequential behavior embeddings, we experiment on a real-world telecommunication dataset with prediction and classification tasks based on the learned embeddings. The experimental results show that the learned embeddings can better identify fraudulent behavior. Finally, we visualize the weights of the attention mechanism to provide a rational interpretation of human behavioral patterns.

In recent years, the research of multistatic synthetic aperture radar (SAR) has become a hot spot because of its great potentials in military and civil applications. However, it is inevitably disturbed by electronic countermeasures. In the multistatic SAR image, the false target caused by transmitting deceptive jamming leads to misjudgment of the image information. In this article, a new antideceptive jamming method for multistatic SAR is proposed. This method can locate the deceptive jammer and effectively suppress it. First of all, the echo model of multistatic SAR in the jamming environment is set up. Based on this model, the relationship expression for the location of jammer is derived. Then, we use the maximally stable extremal region method and the Euclidean-distance-based method to detect and identify false target in the multistatic SAR image. By using the location information of false target, receivers, and transmitters, the expression of the jammer localization is solved to locate the jammer. Finally, the minimum variance distortionless response beamforming method and image mosaic method are used to effectively suppress the jamming. The simulation results show that the method is effective.

Geosynchronous (GEO) spaceborne–airborne bistatic synthetic aperture radar (GEO-BiSAR), where the high-altitude transmitter provides continuous illumination for the receiver, is capable of providing benefits to remote sensing applications. However, obtaining a focused image with high efficiency is a challenging work. The severe 2-D space-variant range cell migration and Doppler modulation introduced by the two moving platforms make the echo hard to be focused. Moreover, the azimuth ambiguity due to the low pulse repetition frequency adopted by the GEO illuminator seriously deteriorates the quality of the final image. In order to simultaneously suppress the azimuth ambiguity and obtain well-focused images, a weighted fast factorized backprojection (FFBP) method is proposed for multichannel GEO BiSAR in this letter. First, the Doppler ambiguity of GEO BiSAR is analyzed and the azimuth multichannel receiving technique is introduced. Then, the multichannel transfer function for GEO BiSAR is derived. Based on the multichannel transfer function, a weighted FFBP method is proposed to achieve accurate focusing and ambiguity suppression. Finally, the simulation results validate the effectiveness of the proposed method.

Synthetic aperture radar (SAR) is an advanced microwave imaging system of great importance. The recognition of real-world targets from SAR images, i.e., automatic target recognition (ATR), is an attractive but challenging issue. The majority of existing SAR ATR methods are designed for single-view SAR images. However, multiview SAR images contain more abundant classification information than single-view SAR images, which benefits automatic target classification and recognition. This paper proposes an end-to-end deep feature extraction and fusion network (FEF-Net) that can effectively exploit recognition information from multiview SAR images and can boost the target recognition performance. The proposed FEF-Net is based on a multiple-input network structure with some distinct and useful learning modules, such as deformable convolution and squeeze-and-excitation (SE). Multiview recognition information can be effectively extracted and fused with these modules. Therefore, excellent multiview SAR target recognition performance can be achieved by the proposed FEF-Net. The superiority of the proposed FEF-Net was validated based on experiments with the moving and stationary target acquisition and recognition (MSTAR) dataset.

Video synthetic aperture radar (SAR) have received more and more attention in recent years as it can provide continuous images of the observed area. However, the enormous data generated by the multi-frame images in video SAR brings big challenges to its transmission, storage and processing, especially for small unmanned aerial vehicle (UAV) platform. In this paper, we aim at proposing an efficient video formation method for SAR system with reduced data. The video formation problem is modelled as a joint low-rank and sparse tensors recovery problem. After that, this problem is solved by an efficient tensor recovery method based on alternating direction method of multiplier. Compared with frequency-domain or time-domain imaging methods, the amount of data samples used can be greatly reduced. Numerical simulations validate the effectiveness of the proposed method.

Synthetic Aperture Radar (SAR) image is highly useful in civilian and military fields, such as ship detection, maritime search and rescue. Considering the target detection from SAR image, we propose a SAR image reconstruction and target extraction method via low-rank and sparsity constrains from under-sampled data. Firstly, the low-rank and sparsity constrains are incorporated into the SAR image reconstruction model, and the objective function is established based on Robust Principal Component Analysis (RPCA) theory. Then, the Augment Lagrange Multiplier (ALM) algorithm is used to transform this objective function to a convex optimization problem. Lastly, SAR image reconstruction and target extraction are obtained by Alternating Direction Method of Multipliers (ADMM) algorithm. The simulations are conducted to verify the effectiveness of the proposed method.

The clutter background poses a challenge to the detection and recognition of targets from synthetic aperture radar (SAR) images, especially when the target is submerged by the clutter. In this paper, we propose a target-oriented SAR waveform optimization method to deal with this problem. The proposed method constructs an optimization criterion jointing the signal-to-clutter ratio (SCR) and the resolution of transmitted waveform. Based on the prior information of the frequency response of the interested target, the clutter suppression performance and the range resolution performance are jointly optimized. The simulation results show that the proposed method can effectively improve the SCR of SAR image in the condition of low SCR, while the resolution performance is guaranteed.

Traditional synthetic aperture radar (SAR) imaging methods focus on high-resolution imaging of the observation scene. In the application of specific target imaging and detection, such as threat target search, the priori knowledge of target can be used for better task performance. Therefore, it is highly desirable to improve the image quality of the target. In this paper, we propose a SAR formation method based on the sparse dictionary. Firstly, the sparse dictionary is learned through the SAR images of a specific target via K-SVD method. Then the SAR formation model is established as a sparse reconstruction problem by incorporating the sparse dictionary. Lastly, the problem is solved via Augment Lagrange Multiplier (ALM) method and Alternating Direction Method of Multipliers (ADMM) method. The proposed method is validated by the simulation, and the results show that the proposed method can effectively reconstruct the target.

Multistatic SAR can obtain information from different angles simultaneously, it has application potential in many fields. However, it still be affected by jammers. When deceptive jammers jamming the SAR system, the false target is generated in the SAR image. In this paper, when the false target is generated on the sea by jammer which located at the coast, the anti-jamming method is proposed. Firstly, echo model of multistatic SAR in deceptive jamming environment is established. Then, false targets are detected and recognized. Next, the jammer localization method is proposed according to relationship between the jammer and false targets. Finally, the jammer location is obtained and jamming signal suppression is achieved. Simulation results validate the effectiveness of the proposed method in this paper.

Multiview synthetic aperture radar (SAR) images contain much richer information for automatic target recognition (ATR) than a single-view one. It is desirable to establish a reasonable multiview ATR scheme and design effective ATR algorithm to thoroughly learn and extract that classification information, so that superior SAR ATR performance can be achieved. Hence, a general processing framework applicable for a multiview SAR ATR pattern is first given in this paper, which can provide an effective approach to ATR system design. Then, a new ATR method using a multiview deep feature learning network is designed based on the proposed multiview ATR framework. The proposed neural network is with a multiple input parallel topology and some distinct deep feature learning modules, with which significant classification features, the intra-view and inter-view features existing in the input multiview SAR images, will be learned simultaneously and thoroughly. Therefore, the proposed multiview deep feature learning network can achieve an excellent SAR ATR performance. Experimental results have shown the superiorities of the proposed multiview SAR ATR method under various operating conditions.

In the era of big data, it is extremely challenging to decide what information to receive and filter out in order to effectively acquire high-quality information, particularly in social media where large-scale User Generated Contents (UGC) is widely and quickly disseminated. Considering that each individual user in social network can take actions to drive the process of information diffusion, it is naturally appealing to aggregate spreading information effectively at the individual level by regarding each user as a social sensor. Along this line, in this paper, we propose a framework for effective information acquisition in social media. To be more specific, we introduce a novel measurement, the preference-based Detection Ability to evaluate the ability of social sensors to detect diffusing events, and the problem of effective information acquisition is then reduced to achieving social sensing maximization through discovering valid social sensors. In pursuit of social sensing maximization, we propose two algorithms to resolve the longstanding problems in traditional greedy methods from the perspectives of efficiency and performance. On the one hand, we propose an efficient algorithm termed LeCELF, which resolves the redundant re-evaluations in the traditional Cost-Effective Lazy Forward (CELF) algorithm. On the other hand, we observe the participation paradox phenomenon in the social sensing network, and proceed to propose a randomized selection-based algorithm called FRIENDOM to choose social sensors to improve the effectiveness of information acquisition. Experiments on a disease spreading network and real-world microblog datasets have validated that LeCELF greatly reduces the running time, whereas FRIENDOM achieves a better detection performance. The proposed framework and corresponding algorithms can be applicable in many other settings in resolving information overload problems.

Despite that the accuracy and efficiency of stereo matching technology have significantly improved in the past decades, the issue of edge-blurring remains a challenge to most of the existing approaches. In this paper, we propose a minimum spanning tree (MST) based stereo matching method by using the image edge and segmentation optimization to preserve the image boundary. We first exploit a fast disparity range estimation method by combining the Surf and Akaze feature points to improve the computational efficiency. Second, we utilize the image edges and brightness information to generate a self-adaptive weight function, which is able to significantly improve the accuracy of MST aggregating in the regions of complex texture and boundaries with similar color distribution. Third, we employ the image segmentation to extract the invalid regions of the estimated disparity map, and propose a post-processing scheme to refine the disparity result. Finally, we run our method on several Middlebury and KITTI datasets. The comparison results between our method and other state-of-the-art approaches demonstrate that the proposed method has high accuracy for disparity computation and is especially robust to the edge-blurring.

Recently, compressive sensing (CS) has been applied in synthetic aperture radar (SAR) imaging, which is increasingly in the focus of study and shows great potential. In CS-based SAR imaging, motion errors of the moving platform introduce inaccuracies in the observation model, which cause various degradations in the final images. To accomplish accurate motion compensation during CS-based SAR imaging, we propose a fast CS-based SAR imaging integrated with motion compensation method. First, CS-based imaging based on the utilization of inverse observation deduced from the inversion of conventional imaging procedures is applied, which is much more computational efficient than the exact observation model. Then, an improved inverse observation model integrated with motion compensation is derived. In the improved model, spatially variant azimuth phase errors are taken into consideration. Joint SAR imaging and motion compensation are formulated as a sparse recovery problem and solved in an iterative way, wherein each iteration both image formation and motion compensation are carried out. The processing of SAR data shows that the proposed method can obtain better focused images compared with the existing SAR imaging and motion compensation methods.

The cross-lingual sentiment analysis (CLSA) aims to leverage label-rich resources in the source language to improve the models of a resource-scarce domain in the target language, where monolingual approaches based on machine learning usually suffer from the unavailability of sentiment knowledge. Recently, the transfer learning paradigm that can transfer sentiment knowledge from resource-rich languages, for example, English, to resource-poor languages, for example, Chinese, has gained particular interest. Along this line, in this article, we propose semisupervised learning with SCL and space transfer (ssSCL-ST), a semisupervised transfer learning approach that makes use of structural correspondence learning as well as space transfer for cross-lingual sentiment analysis. The key idea behind ssSCL-ST, at a high level, is to explore the intrinsic sentiment knowledge in the target-lingual domain and to reduce the loss of valuable knowledge due to the knowledge transfer via semisupervised learning. ssSCL-ST also features in pivot set extension and space transfer, which helps to enhance the efficiency of knowledge transfer and improve the classification accuracy in the target language domain. Extensive experimental results demonstrate the superiority of ssSCL-ST to the state-of-the-art approaches without using any parallel corpora.

Multi-agent reinforcement learning (MARL) has attracted more and more attention in recent years. It is now widely applied in various fields, including cyber physical systems, smart grid, finance, social network, and among others. The current researches on MARL mainly focus single-time scale, in which the agents have the same decision epoch. While in real applications, it is common that the agents make decisions by different frequencies. In addition, different agents may have separate roles in the system. In this paper, we propose a more general MARL framework by introducing multi-time scale of decision epochs. We assume that agents share information with their neighbors, including state, action, and reward. The global observability of state and action, which is a common assumption, is not required. We propose a decentralized Q-learning algorithm and a modified MADDPG algorithm to solve the problem. The main contributions of this paper are as follows. First, we formulate the multi-time scale multi-agent reinforcement learning (MTMARL) problem. This provides a general framework for the related systems and problems. Second, we provide a networked decentralized multi-time scale multi-agent Q-learning algorithm to solve the problem and prove its convergence. Third, we test the algorithm numerically. The results show that the proposed algorithm performs better than the previous QD-learning and is only slightly worse than the centralized algorithm.

The recognition accuracy of micro-expressions in the field of facial expressions is still understudied, as current research methods mainly focus on feature extraction and classification. Based on optical flow and decision thinking theory, we propose a novel micro-expression recognition method, which can filter low-quality micro-expression video clips. Determined by preset thresholds, we develop two optical flow filtering mechanisms: one based on two-branch decisions (OFF2BD) and the other based on three-way decisions (OFF3WD). In OFF2BD, which use the classical binary logic to classify images, and divide the images into positive or negative domain for further filtering. Differ from the OFF2BD, OFF3WD added boundary domain to delay to judge the motion quality of the images. In this way, the video clips with low degree of morphological change can be eliminated, so as to directly improve the quality of micro-expression features and recognition rate. From the experimental results, we verify the recognition accuracy of 61.57%, and 65.41% for CASMEII, and SMIC datasets, respectively. Through the comparative analysis, it shows that the scheme can effectively improve the recognition performance.

This paper presents a long-time integration method for passive radar detection of marine target with multi-stage motions using global navigation satellite system (GNSS) as illuminator. Owing to the complex motion, the range cell migration (RCM) and the Doppler parameters of the target echo pertaining the multiple motion stages are different. To solve the problems, in this method, first the range symmetry transform (RST) is applied to correct the RCM over the multiple stages, after which a one-dimensional azimuth signal can be directly extracted from the two-dimensional echo. Then, the conjugate integration processing (CIP) is performed to accumulate the target energy during each stage into Doppler centroid (DC) and Doppler frequency rate (DFR) domain. Finally, in order to settle the difference in Doppler parameters between multiple stages, the DC-DFR maps are projected and combined in high-dimensional Doppler parameter domain after Doppler frequency shift compensation, implementing the well integration of the echo over the entire long time. The effectiveness of the proposed integration method is demonstrated by simulations.

Deep learning (DL)-based auto-segmentation has the potential for accurate organ delineation in radiotherapy applications but requires large amounts of clean labeled data to train a robust model. However, annotating medical images is extremely time-consuming and requires clinical expertise, especially for segmentation that demands voxel-wise labels. On the other hand, medical images without annotations are abundant and highly accessible. To alleviate the influence of the limited number of clean labels, we propose a weakly supervised DL training approach using deformable image registration (DIR)-based annotations, leveraging the abundance of unlabeled data. We generate pseudo-contours by utilizing DIR to propagate atlas contours onto abundant unlabeled images and train a robust DL-based segmentation model. With 10 labeled TCIA dataset and 50 unlabeled CT scans from our institution, our model achieved Dice similarity coefficient of 87.9%, 73.4%, 73.4%, 63.2% and 61.0% on mandible, left & right parotid glands and left & right submandibular glands of TCIA test set and competitive performance on our institutional clinical dataset and a third party (PDDCA) dataset. Experimental results demonstrated the proposed method outperformed traditional multi-atlas DIR methods and fully supervised limited data training and is promising for DL-based medical image segmentation application with limited annotated data.

Cross-lingual sentiment classification (CLSC) aims to leverage rich-labeled resources in the source language to improve prediction models of a resource-scarce domain in the target language. Existing feature representation learning-based approaches try to minimize the difference of latent features between different domains by exact alignment, which is achieved by either one-to-one topic alignment or matrix projection. Exact alignment, however, restricts the representation flexibility and further degrades the model performances on CLSC tasks if the distribution difference between two language domains is large. On the other hand, most previous studies proposed document-level models or ignored sentiment polarities of topics that might lead to insufficient learning of latent features. To solve the abovementioned problems, we propose a coarse alignment mechanism to enhance the model’s representation by a group-to-group topic alignment into an aspect-level fine-grained model. First, we propose an unsupervised aspect, opinion, and sentiment unification model (AOS), which trimodels aspects, opinions, and sentiments of reviews from different domains and helps capture more accurate latent feature representation by a coarse alignment mechanism. To further boost AOS, we propose ps-AOS, a partial supervised AOS model, in which labeled source language data help minimize the difference of feature representations between two language domains with the help of logistics regression. Finally, an expectation–maximization framework with Gibbs sampling is then proposed to optimize our model. Extensive experiments on various multilingual product review data sets show that ps-AOS significantly outperforms various kinds of state-of-the-art baselines.

Partly due to the use of exhaustive-annotated data, deep networks have achieved impressive performance on medical image segmentation. Medical imaging data paired with noisy annotation are, however, ubiquitous, but little is known about the effect of noisy annotation on deep learning based medical image segmentation. We studied the effect of noisy annotation in the context of mandible segmentation from CT images. First, 202 images of head and neck cancer patients were collected from our clinical database, where the organs-at-risk were annotated by one of twelve planning dosimetrists. The mandibles were roughly annotated as the planning avoiding structure. Then, mandible labels were checked and corrected by a head and neck specialist to get the reference standard. At last, by varying the ratios of noisy labels in the training set, deep networks were trained and tested for mandible segmentation. The trained models were further tested on other two public datasets. Experimental results indicated that the network trained with noisy labels had worse segmentation than that trained with reference standard, and in general, fewer noisy labels led to better performance. When using 20% or less noisy cases for training, no significant difference was found on the segmentation results between the models trained by noisy or reference annotation. Cross-dataset validation results verified that the models trained with noisy data achieved competitive performance to that trained with reference standard. This study suggests that the involved network is robust to noisy annotation to some extent in mandible segmentation from CT images. It also highlights the importance of labeling quality in deep learning. In the future work, extra attention should be paid to how to utilize a small number of reference standard samples to improve the performance of deep learning with noisy annotation.

With the increasing use of recommender systems in various application domains, many algorithms have been proposed for improving the accuracy of recommendations. Among a number of other dimensions of recommender systems performance, long-tail (niche) recommendation performance remains an important challenge, due in large part to the popularity bias of many existing recommendation techniques. In this study, we propose CORE, a cosine-pattern-based technique, for effective long-tail recommendation. Comprehensive experimental results compare the proposed approach to a wide variety of classical, widely-used recommendation algorithms and demonstrate its practical benefits in accuracy, flexibility, and scalability, in addition to the superior long-tail recommendation performance.

More attention is paid for the feature importance ranking (FIR), in particular when high-throughput features can be extracted for intelligent diagnosis and personalized medicine. A large number of FIR methods have been proposed, while few are integrated for comparison and real-life applications. In this study, a matlab toolbox is presented and a total of 30 algorithms are collected. Moreover, the toolbox is evaluated on a database of 163 ultrasound images. To each breast lesion, 15 features are handcrafted. And to Figure out an optimal subset of features for classification, all combinations of features are tested and linear support vector machine is used for the malignancy prediction of lesions annotated in ultrasound images. At last, the effectiveness of FIR is analyzed according to performance comparison. The toolbox is available (https://github.com/NicoYuCN/matFIR). In the future work, more FIR methods, feature selection methods and machine learning classifiers will be integrated.

Radar high resolution imaging becomes a research hotspot in recent years. Synthetic Aperture Radar (SAR) uses motion of single platform to obtain large aperture. However, it can only obtain a limited resolution at a specific perspective. Distributed radar, which utilize platforms to expand in space and form a larger aperture. It records echo from multiple signal channels which can achieve high resolution in all directions. The existing distributed radar system mainly realizes the imaging of static target with lots of platforms. In this paper, we studied moving target imaging of distributed radar in sparse array condition. In this system, all the array elements can work flexibly as transmitter or receiver. Some methods of spectrum estimation are used to solve distributed radar imaging model. But they are all applied to static target and it is difficult to estimate the number of scattered points accurately, especially for moving target. We established the signal model of moving target in a period of observation time and solved it by using truncated singular value decomposition (TSVD) method. Simulation results verified the feasibility of the imaging model and excellent performance of proposed method.

Reliable automatic ship detection in Synthetic Aperture Radar (SAR) image plays an important role in the surveillance of maritime activity. But, the coastal background has seriously affected the accuracy and speed of ship detection. Within this paper, we present a Dual Downscaling Spectral Residual (DDSR) method of sea-land segmentation-based spectral residual for ship detection that attempts to combine the downscaling thresholding operator, saliency computation and morphological features. For this, sea-land segmentation based on the downscaling thresholding operator and morphological features of the potential detection result is applied, which is followed by downscaling Spectral Residual detector so that the accurate detection result is obtained. Measured data illustrate that the proposed method obtain outstanding performance in ship targets detection.

With appropriate geometry configurations, bistatic synthetic aperture radar (SAR) can break through the limitations of monostatic SAR on forward-looking imaging. Thanks to such a capability, bistatic forward-looking SAR (BFSAR) has extensive potential applications. Thus, moving target detection (MTD) for BFSAR has become an important research field. In this paper, a MTD method based on BP-DPCA in BFSAR is presented, and do the detailed analysis. First, back-projection can completely compensate the phase of static clutter, and eliminate the effects of channel spacing for static clutter, but for moving target, the phases can not be compensated fully. Thus, the BP-DPCA method can suppress the static clutter, achieve the goal of moving target detection. Simulation results verify the validity and efficiency of the presented method.

Ground Moving Target Indication (GMTI) for Near-space Hypersonic vehicle borne Synthetic Aperture Radar (NH-SAR) is an important part of remote sensing applications. However, the ultra-long range and the ultra-high flight speed of NH-SAR causes the lower echo signal-to-noise ratio (SNR) and the expanded clutter spectrum, which reduces the performance of traditional GMTI methods such as Displaced Phase Centre Antennas (DPCA), along track interferometry (ATI) and Space Time Adaptive Processing (STAP). In this paper, a method based on Track Before Detect (TBD) in image domain is proposed to perform GMTI for NH-SAR. This method obtains the multi-frame images of NH-SAR firstly, and tracks the trajectories of the moving targets in the images. The simulation results validate the effectiveness of the proposed method, and show that this method has lower false alarm probabilities under the low-SNR condition.

During the inshore ship detection in SAR images, the high similarity between the harbor and the ship body on gray and texture features, resulting in low detection accuracy and high false alarm rate. In this paper, we propose a new approach using contextual fluctuation information to deal with this problem. Firstly, the maximum stability extremal region (MSER) method is used for global pre-screening to quickly obtain candidate targets. Next, the context slice of each candidate target is obtained, then each slice is devided into grids. Distinguish real ship targets from false alarms based on the fluctuations in pixel values within the divided grid. Experimental results based on satellite-borne SAR data illustrate that the proposed method obtains excellent detection performance and low false alarm rate.

Intelligence technology has been widely used in various radar applications in recent years. However, there are few studies on the combination of intelligence technology and synthetic aperture radar (SAR). In this paper, a cognitive SAR waveform design method based on joint optimization criteria is proposed. The proposed method constructs a criteria jointing the resolution of SAR transmitting waveform and output SCNR. Cognitive SAR waveform designed upon this criterion solves the problem of weak clutter suppression and poor cognitive ability of conventional SAR. Meanwhile, it also solves the problem that the resolution performance of waveform designed based on the traditional criterion of maximum output SCNR is too poor to be used in SAR for high resolution missions. Simulation results show that the waveform designed by this method has high SCNR and great resolution.

The constant false alarm rate with convolution and pooling (CP-CFAR) method, which can improve the detection efficiency via GPU parallel acceleration in the airborne synthetic aperture radar (SAR) images, is proposed in this paper. The constant false alarm rate (CFAR) method is one of the most widely used methods for target detection in airborne SAR images. However, since the CFAR is performed on a pixel-by-pixel basis, the time consumption will increase rapidly with the expansion of image scene. Even if the GPU is used for acceleration, the efficiency improvement is still limited, which cannot meet the real-time processing requirements. Therefore, the CP-CFAR method for the target detection of SAR images is proposed in this paper. The convolution layer uses the horizontal and vertical Sobel operators to improve the contrast between targets and background, and the pooling layer can reduce the processing dimension of the images. The convolution and pooling layers are added before the two-parameter CFAR, which can reduce the computational elements but without losing the main feature of the original image. More importantly, compared to the traditional CFAR, the proposed CP-CFAR is more suitable for GPU acceleration, which can improve the detection efficiency significantly. Experiments on the moving and stationary target acquisition and recognition SAR images with a size of 1478 × 1784 show that, compared with the traditional cell-averaging CFAR, two-parameter CFAR and their CPU, multithread CPU, and GPU acceleration modes, the proposed CP-CFAR with GPU acceleration can obtain the best detection performance with the highest acceleration ratio, and the operation time is less than 192 ms.

Consensus clustering aims to find a single partition of data that agrees as much as possible with existing basic partitions. Given its robustness and generalizability, consensus clustering has emerged as a promising solution to find cluster structures inside heterogeneous big data rising from various application domains. In the area of fuzzy systems, however, research along this line is still in its initial stage with some unsystematic algorithmic studies. Finding a fuzzy consensus partition from multiple fuzzy basic partitions in an efficient, flexible, and robust way is still an exciting open problem calling for further investigation. In light of this, this paper provides a systematic study of fuzzy consensus clustering (FCC) from a utility perspective. Specifically, we first define the objective function of FCC clearly using the novel fuzzified contingency matrix. We then derive a family of FCC Utility functions termed as FCCU that can transform FCC to a weighted piecewise fuzzy $c$ -means clustering (piFCM) problem. This helps us to establish an algorithmic framework for FCC with flexible choice of utility functions, and speeds FCC significantly with a FCM-like iterative process of piFCM. To meet the big data challenge, we further parallelize FCC on the Spark platform with both vertical and horizontal segmentation schemes. Extensive experiments on various real-world datasets demonstrate the excellent performance of FCC, even with a majority of poor basic partitions. In particular, our method exhibits interesting potential for big data clustering in two real-life applications concerned with online event detection and overlapping community detection, respectively.

Multiband acquisition, also called simultaneous multislice, has become a popular technique in resting-state functional connectivity studies. Multiband (MB) acceleration leads to a higher temporal resolution but also leads to spatially heterogeneous noise amplification, suggesting the costs may be greater in areas such as the subcortex. We evaluate MB factors of 2, 3, 4, 6, 8, 9, and 12 with 2 mm isotropic voxels, and additionally 2 mm and 3.3 mm single-band acquisitions, on a 32-channel head coil. Noise amplification was greater in deeper brain regions, including subcortical regions. Correlations were attenuated by noise amplification, which resulted in spatially varying biases that were more severe at higher MB factors. Temporal filtering decreased spatial biases in correlations due to noise amplification, but also tended to decrease effect sizes. In seed-based correlation maps, left-right putamen connectivity and thalamo-motor connectivity were highest in the single-band 3.3 mm protocol. In correlation matrices, MB 4, 6, and 8 had a greater number of significant correlations than the other acquisitions (both with and without temporal filtering). We recommend single-band 3.3 mm for seed-based subcortical analyses, and MB 4 provides a reasonable balance for studies analyzing both seed-based correlation maps and connectivity matrices. In multiband studies including secondary analyses of large-scale datasets, we recommend reporting effect sizes or test statistics instead of correlations. If correlations are reported, temporal filtering (or another method for thermal noise removal) should be used. The Emory Multiband Dataset is available on OpenNeuro.

In spite of the fact that convolutional neural network-based stereo matching models have shown good performance in both accuracy and robustness, the issue of image feature loss in regions of texture-less, complex scenes and occlusions remains. In this paper, we present a dense convolutional neural network-based stereo matching method with multiscale feature connection, named Dense-CNN. First, we construct a novel densely connected network with multiscale convolutional layers to extract rich image features, in which the merged multiscale features with context information are utilized to estimate the cost volume for stereo matching. Second, we plan a novel loss-function strategy to learn the network parameters more reasonably, which can develop the performance of the proposed Dense-CNN model on disparity computation. Finally, we run our Dense-CNN model on the Middlebury and KITTI databases to conduct a comprehensive comparison with several state-of-the-art approaches. The experimental results demonstrate that the proposed method achieved superior performance on computational accuracy and robustness of disparity estimation, especially achieving the significant benefit of feature preservation in ill-posed regions.

In bistatic forward-looking synthetic aperture radar (BFSAR), conventional autofocus algorithms cannot estimate the phase errors accurately when the range walk is compensated in the azimuthal time domain. This problem stems from the influence of the azimuth-variant Doppler coefficients after linear range cell migration correction in azimuthal time domain. In principle, nonlinear chirp scaling processing can be carried out to remove the azimuth variance. Nevertheless, Doppler azimuth variations of the Doppler parameters induced by the motion errors are not taken into consideration and serious defocus would emerge in the final image. To cope with such a problem, an estimation–evaluation–equalization scheme is proposed before conventional autofocus for BFSAR. Different from the conventional autofocus method, the azimuth-variant Doppler coefficients are additionally estimated and equalized before autofocus, and as a consequence, phase errors can be precisely estimated. The BFSAR data processing results demonstrate the validity of the proposed method on the improvement of autofocus in BFSAR.

Bistatic forward-looking synthetic aperture radar (BFSAR) is a kind of bistatic SAR system that can image forward-looking terrain in the flight direction of a moving platform. In BFSAR, compensation of the flight trajectory errors is of great significance to get a well-focused image. To accomplish an accurate motion compensation in image processing, a high-precision navigation system is needed. However, in many cases, due to the accuracy limit of such systems, flight trajectory errors are hard to be compensated correctly, causing mainly the resolution decrease in final images. In order to cope with such a problem, we propose a rise-dimensional modeling and estimation for flight trajectory error based on raw BFSAR data in this paper. To apply this method, we first carry out a preprocessing named azimuth-slowtime decoupling to deal with the spatially variant flight trajectory error before estimation. Then, an optimization model for flight trajectory estimation under the criterion of maximum image intensity is built. The solution to the optimization model is the accurate flight trajectory. Then, block coordinate descent technique is used to solve this optimization model. The processing of BFSAR data shows that the algorithm can obtain a more accurate estimation results, and generate better focused images compared with the existing trajectory estimation method.

With appropriate geometry configurations, bistatic synthetic aperture radar (SAR) can break through the limitations of monostatic SAR on forward-looking imaging. Owing to such a capability, bistatic forward-looking SAR (BFSAR) has extensive potential applications. In BFSAR, the compensation of the spatially variant motion errors is of great significance to get a well-focused image. In this paper, first, the spatial-variance properties of motion errors are analyzed analytically and quantitatively. Different from the side-looking monostatic and bistatic SAR, 2-D space-variant motion errors should be taken into consideration in BFSAR. The 2-D spatial variance of the motion errors can be categorized into two parts, range-variant motion errors of the transmitter and azimuth-variant motion errors of the receiver. Moreover, these two parts are independent of each other. Based on this property analysis, second, a motion compensation (MoCo) approach with cubic-order processing is proposed to deal with the spatially variant motion errors in BFSAR. In the cubic-order processing, the first-order MoCo is performed to correct the spatially independent motion errors on the raw data. The second-order MoCo is accomplished on the non-range-cell-migration (RCM) data to deal with the range-variant errors. After the second-order MoCo, since the signal direction of the non-RCM data coincides with the variant direction of the uncompensated phase errors, the azimuth-variant motion errors and slow time signal are coupled together. To cope with such a problem, the slow time signal is transformed into the direction perpendicular to the azimuth by a novel procedure named azimuth–slow time decoupling. At this stage, the coupling between the azimuth-variant motion errors and slow time signal has been eliminated. Azimuth-variant motion errors can be corrected precisely. Simulation and experimental results verify the effectiveness of the proposed method.

With the geosynchronous synthetic aperture radar (SAR) satellite as the transmitter, the unmanned aerial vehicle (UAV) can passively receive the echo within the illuminated ground area and achieve 2-D imaging of the interested target. This SAR system, known as GEO-UAV bistatic SAR, is capable of autonomously accomplishing the bistatic SAR mission in rough terrain environments by prespecifying a path for the UAV receiver. In this paper, the GEO-UAV bistatic SAR system is first investigated. The practical advantages and spatial resolution are then analyzed in detail. The spatial resolution of GEO-UAV bistatic SAR is dependent on the observation geometry, which is determined by the UAV path. Therefore, the path planning for GEO-UAV bistatic SAR aims at identifying a set of optimal paths for the UAV receiver to travel through a 3-D terrain environment that simultaneously guarantees the safety of the UAV and achieves SAR imaging with optimized performance during the flight. The path planning is modeled as a constrained multiobjective optimization problem (MOP), which accurately represents the two main aspects for the path planning problem, i.e., UAV navigation and bistatic SAR imaging. Then, a path planning method based on a constrained-adaptive-multiobjective-differential-evolution algorithm is proposed to solve the MOP and generate multiple feasible paths for the UAV receiver with different tradeoffs between navigation for UAV and bistatic SAR imaging performance. The GEO-UAV bistatic SAR mission designer can choose a path from the solution set according to the application requirements, which makes the method more pragmatic.

Multistatic synthetic aperture radar (SAR) have achieved significant performance in the field of anti-jamming for its flexible configuration. Hence, its capacity of anti-jamming is very important in electronic warfare. In addition, it is difficult to remove the false targets caused by the deceptive jammer from SAR echoes. We propose a novel anti-deceptive jamming method for multistatic SAR in this paper which is able to locate the deceptive jammer and eliminate its influence. We first apply maximally stable extremal region (MSER) and European-distances-based method to detect the false targets. The position of the jammer can then be obtained by exploiting the position information of the false targets and the SAR transmitters/receivers. At last, beamforming is applied to achieve the anti-deceptive jamming. Experimental results demonstrate the effectiveness of our proposed method.