Search

The most tedious and hectic job is to identify the tiny benthic animals by spending thousands of hour under the microscope, since all the fauna need to be counted, sorted, picked and permanently mounted on glass slides for taxonomic identification. All faunal identifications need a lot of preprocessing and it consumes a lot of time to identify a single specimen. Therefore, to reduce the complexity of many such procedures, combined with the desire to identify larger datasets, we came up with new software based on artificial intelligence which can automatically identify the benthic fauna through the microscopic images. In this paper, we propose a machine learning method for automatic visual identification through the images of the benthic fauna. To this end, we propose a neural network model, where we demonstrate that the proposed approach differentiates the fauna based on images. However, it works well with vast amounts of image data and significant computational resources.

The grid-based approach is popular for point-in-polygon tests. However, there is a trade-off between the preprocessing and the inclusion test, which always requires the grid resolutions to be tuned. In this article, we address this challenge by enhancing the grid structure using y-axis-aligned stripes, which are formed by the y-axis-aligned lines passing through the endpoints of the edge segments in the cell, thereby managing the edge segments in each grid cell. Moreover, we precompute the inclusion properties of the x-axis-aligned top borders of the stripes during preprocessing. Therefore, to answer a query point with the ray crossing method, we can emit a ray from the point to propagate upwards until the ray arrives at the top border of a stripe. We thoroughly consider singular cases to guarantee each query point can be answered in the stripe that contains the point. In our method, the computational load can be decreased, as one coordinate of the intersection point between the ray and an edge is known in advance, and parallel computing can be well exploited because the branching operations for determining whether an edge intersects with the ray are saved. Experimental results show that the efficiency of our method does not vary much with respect to the grid resolutions, so the trouble of tuning grid resolutions can be avoided. Ultimately, our method with a low grid resolution can reduce the preprocessing time and still achieve a higher inclusion test efficiency than the existing methods with a high grid resolution, especially on GPUs.

To identify the abnormal characteristics of the oplegnathus punctatus is great importance to the detection of iridovirus disease in the breeding environment. In this paper, an advanced neural network model to identify the characteristics of the oplegnathus punctatus and predict its different periods of suffering from iridovirus disease is proposed based on the establishment of a data set. First of all, a standard format data set of oplegnathus punctatus and an abnormal format date set are established in order to verify the effectiveness of the method in this paper. And then, the feature extraction fusion method is used for preprocessing in terms of the abnormal format data set, which combines the edge features extracted by the improved multi-template Sobel operator and the color features extracted by the HSV model. Finally, an improved VGG-GoogleNet network recognition model comes into being through the fusion and improvement of the VGG and GoogleNet neural network structure. The experiments results show that the prediction accuracy rate for oplegnathus punctatus suffering from iridovirus disease in the the abnormal format data set and the standard format data set are improved, which reach 98.55% and 69.18%.

There is a growing interest in the neuroscience community on the advantages of multimodal neuroimaging modalities. Functional and structural interactions between brain areas can be represented as a network (graph) allowing us to employ graph-theoretic tools in multiple research directions. Researchers usually treated brain networks acquired from different modalities or different frequencies separately. However, there is strong evidence that these networks share complementary information while their interdependencies could reveal novel findings. For this purpose, neuroscientists adopt multilayer networks, which can be described mathematically as an extension of trivial single-layer networks. Multilayer networks have become popular in neuroscience due to their advantage to integrate different sources of information. We can incorporate this information from different modalities (multi-modal case), from different frequencies (multi-frequency case), or a single modality following a dynamic functional connectivity analysis (multi-layer, dynamic case). Researchers already used multi-layer networks to model brain disorders, to detect key hubs related to a specific function, to reveal structural-functional relationships, and to define more precise connectomic biomarkers related to brain disorders. However, the construction of a multilayer network depends on the selection of multiple preprocessing steps that can affect the final network topology. Here, we analyzed the fMRI dataset from a single human performing scanning over a period of 18 months (84 scans in total). We focused on assessing the reproducibility of multi-frequency multilayer topologies exploring the effect of two filtering methods for extracting frequencies from BOLD activity, three connectivity estimators, with or without a topological filtering scheme, and two spatial scales. Finally, we untangled specific combinations of researchers choices that yield repeatable topologies, giving us the chance to recommend best practices over consistent topologies.

The driver's fatigue directly affects the safety factor of the compact vehicle driving in actual road. Mastering the driver's fatigue state plays an important role in the driver's safety driving and timely adjustment of mental state. In view of the particularity of the driving safety of the compact vehicle, this paper takes the driver's brain electricity (EEG) signal as the research object, and starts from the formulation of the experimental scheme, and based on the special training system in the simulation driving software. Two types of driving quality evaluation indicators: the fine operation ability and emergency response capability is formulated; after preprocessing and eigenvalue selection of EEG signals, DPCA clustering algorithm combined with driving quality is used to complete the classification of driver fatigue and the marking of EEG signal feature data set. Finally, the driver fatigue recognition model is initially constructed by using the labeled data set combined with the convolutional neural network (CNN).

To avoid unexpected failures of engineering systems, sensors have been widely used to monitor the degradation process of the systems. A number of studies have been conducted to analyze the collected sensor signals and predict the failure time. However, the existing studies are usually restricted and cannot be adapted to different practical situations. In this paper, we propose a systematic method for degradation modeling and prognosis that can be widely applied in different scenarios. In particular, the proposed method is capable to handle one or multiple sensors, powerful to capture the nonlinear relations between sensor signals and the degradation process with few assumptions, generic to consider multiple failure modes, flexible to deal with unequally spaced sensor measurements or asynchronous signals, and easily understandable with little preprocessing required. The main idea is to predict the failure time of an in-service unit based on a subset of the nearest historical units, where features are extracted from each sensor to describe the progression of sensor signals and local linear regression models are constructed to establish the relation between failure time and the extracted features. The prediction variance is then used as the goodness-of-fit measure, based on which decision-level fusion and feature-level fusion are proposed to combine multiple sensors. A case study with two datasets on the degradation modeling of aircraft engines is conducted which shows satisfactory performance of the proposed method.

Retrieval of mathematical information from scientific documents is one of the crucial tasks. Numerous Mathematical Information Retrieval (MIR) systems have been developed, which mainly focus on the improvement over the indexing and the searching mechanism, the poor results obtained for evaluation measures depict major limitations of such systems. These enhance the scope of improvement and new innovations through the inclusion of functionalities, which can resolve the challenges of MIR system. Further, to improve the performance of the MIR systems, this paper proposed a formula embedding and generalization approach with the context, in addition to innovative relevance measurement technique. In this approach, documents are preprocessed by the document preprocessor module and extracted the formulas in Presentation MathML format with their context. The formula embedding and generalization modules of the proposed approach formed the binary vectors where ‘1’ represents the presence, and ‘0’ represents the absence of a particular entity in a formula, and subsequently, the vectors of formulas with context are indexed by the indexer. The innovative relevance measurement technique of the proposed approach ranked those documents first, which are retrieved by both formula embedding and generalization modules as compared to the individual one. The proposed approach has been tested on the MathTagArticles of Wikipedia of NTCIR-12, and the obtained results verify the significance of the context of the formula and the dissimilarity factor in the retrieval of mathematical information.

Offensive communications have invaded social media content. One of the most effective solutions to cope with this problem is using computational techniques to discriminate offensive content. Moreover, social media users are from linguistically different communities. This study aims to tackle the Multilingual Offensive Language Detection (MOLD) task using transfer learning models and the fine-tuning phase. We propose an effective approach based on the Bidirectional Encoder Representations from Transformers (BERT) that has shown great potential in capturing the semantics and contextual information within texts. The proposed system consists of several stages: (1) Preprocessing, (2) Text representation using BERT models, and (3) Classification into two categories: Offensive and non-offensive. To handle multilingualism, we explore different techniques such as the joint-multilingual and translation-based ones. The first consists in developing one classification system for different languages, and the second involves the translation phase to transform all texts into one universal language then classify them. We conduct several experiments on a bilingual dataset extracted from the Semi-supervised Offensive Language Identification Dataset (SOLID). The experimental findings show that the translation-based method in conjunction with Arabic BERT (AraBERT) achieves over 93% and 91% in terms of F1-score and accuracy, respectively.

Machine learning and its subfield deep learning techniques provide opportunities for the development of operator mental state monitoring, especially for cognitive workload recognition using electroencephalogram (EEG) signals. Although a variety of machine learning methods have been proposed for recognizing cognitive workload via EEG recently, there does not yet exist a review that covers in-depth the application of machine learning methods. To alleviate this gap, in this paper, we survey cognitive workload and machine learning literature to identify the approaches and highlight the primary advances. To be specific, we first introduce the concepts of cognitive workload and machine learning. Then, we discuss the steps of classical machine learning for cognitive workload recognition from the following aspects, i.e., EEG data preprocessing, feature extraction and selection, classification method, and evaluation methods. Further, we review the commonly used deep learning methods for this domain. Finally, we expound on the open problem and future outlooks.

In this paper, we propose an improvement in the preprocessing stage of the reversible data hiding method based on contrast enhancement applied to grayscale digital images. In conventional proposals, the preprocessing stage plays a fundamental role in the performance of the data hiding algorithm in terms of capacity and visual quality. Thus, this work presents two improvements to the preprocessing stage based on the analysis, merging, and shifting the image histogram, which, in conjunction with the data hiding stage, maximize the embedding capacity while keeping the visual distortion at a minimum. The proposed algorithm was implemented on two image datasets and compared with several related works previously reported in the scientific literature. The experimental results show that the visual quality can be preserved after data have been concealed when employing different embedding rates. Either of the preprocessing techniques presented in this paper can be adopted considering the application scenario and the priority of the requirements of the capacity and visual quality needed.

In e-commerce, online retailers are usually suffering from professional malicious users (PMUs), who utilize negative reviews and low ratings to their consumed products on purpose to threaten the retailers for illegal profits. Specifically, there are three challenges for PMU detection: 1) professional malicious users do not conduct any abnormal or illegal interactions (they never concurrently leave too many negative reviews and low ratings at the same time), and they conduct masking strategies to disguise themselves. Therefore, conventional outlier detection methods are confused by their masking strategies. 2) the PMU detection model should take both ratings and reviews into consideration, which makes PMU detection a multi-modal problem. 3) there are no datasets with labels for professional malicious users in public, which makes PMU detection an unsupervised learning problem. To this end, we propose an unsupervised multi-modal learning model: MMD, which employs Metric learning for professional Malicious users Detection with both ratings and reviews. MMD first utilizes a modified RNN to project the informational review into a sentiment score, which jointly considers the ratings and reviews. Then professional malicious user profiling (MUP) is proposed to catch the sentiment gap between sentiment scores and ratings. MUP filters the users and builds a candidate PMU set. We apply a metric learning-based clustering to learn a proper metric matrix for PMU detection. Finally, we can utilize this metric and labeled users to detect PMUs. Specifically, we apply the attention mechanism in metric learning to improve the model's performance. The extensive experiments in four datasets demonstrate that our proposed method can solve this unsupervised detection problem. Moreover, the performance of the state-of-the-art recommender models is enhanced by taking MMD as a preprocessing stage., Comment: Accepted by IEEE Transactions on Knowledge and Data Engineering (TKDE)

Robotic 3D measurement for evaluating the final shape of blades can improve product quality and reduce fabrication costs. However, large or complex objects require multiple scans from different viewpoints to capture the entire object. Errors in registration artificially introduce defects and surface discontinuties in the reconstructed object model. To address the sources of inaccurate registration for methods relying on pairwise associated points, this manuscript describes an inlier selection approach based on the geometric properties of rigid registration, called RiGID. It readily applies to registration strategies that use 3D interest points and descriptors to associate data across overlapping point clouds. RiGID converts geometric constraint violations into distributions over geometric error values. Analysis of the distributions provides a deterministic and data adaptive way to identify inlier sets. RiGID is an efficient preprocessor to rigid registration. It reduces the amount of input data and improves the inlier ratio, which translates to more efficient and accurate registration. Evaluation of the method shows a higher inlier rate and lower registration error compared to baseline outlier rejection methods. The approach is well suited to surfaces with smooth regions and a high percentage of outliers (up to 90\%), which engine and turbine blades exhibit.

Recovering the three-dimensional structure information from a monocular camera is significant for automated driving, robot navigation, and traffic safety assessment. Recent work has solved various tight issues on self-supervised monocular depth estimation of leveraging on-board videos, such as occlusion/disocclusion, dynamic objects, and scale inconsistent. Nevertheless, rare work focuses on the model's prediction confidence and underlying relations between depths, while they are essential for a decision-making system and performance improvement, respectively. This paper proposes a novel scheme, that of correlation-aware structure, to dig into the relations between depths, converting the independent depths into a graph-like connected depth map. Subsequently, a Gaussian estimator is devised to predict the depth map and uncertainty map concurrently. The uncertainty map can show us problematic regions where it is difficult to predict, from which we further develop uncertainty-based strategies to improve the performance. Specifically, we propose a simple image preprocessing method to overcome the gradient locality issue caused by low-texture, especially in smooth roads and shadows. Also, to avoid the influence of high-uncertainty regions, we propose a solidity-aware mask to recognize the reliable pixels for training in the image. The experiments on the KITTI dataset show that our method results in a competitive performance in both depth and ego-motion estimation tasks compared with the state-of-the-art methods. Besides, additional experiments on the Make3D and Cityscapes datasets demonstrate our method's strong generalization capability and practicality.

As the ocean engineering structure serves in the critical underwater environment, a variety of defects, such as cracks and corrosions, always cause damage to the structure. It is still a key challenge to identify and evaluate these defects accurately in the underwater environment. In this paper, the visual and intelligent identification methods are presented for the inspection of defects in underwater structures using the alternating current field measurement (ACFM) technique. The current perturbation theory is developed to analyze the disturbed current field and the distorted magnetic field caused by defects. The gradient imaging algorithm is presented as an image preprocessing method to highlight the visual morphology of defects. The underwater intelligent ACFM system is set up. The experiments are carried out to verify the gradient imaging algorithm. The convolutional neural network (CNN) deep learning algorithm is presented to identify the grey-scale map samples preprocessed by the gradient imaging algorithm. The results show that the current perturbation theory clarifies the relationship between the characteristic signal and the morphology of various defects. The Bz image reflects the surface morphology of defects. The gradient imaging algorithm can achieve visual detection of defects. The single crack, the irregular crack and the corrosion can be identified intelligently by the CNN deep learning algorithm. These defects can be evaluated accurately after classification.

According to the International Agency for Research on Cancer (IARC), the mortality rate due to brain tumors is 76%. It is required to detect the brain tumors as early as possible and to provide the patient with the required treatment to avoid any fatal situation. With the recent advancement in technology, it is possible to automatically detect the tumor from images such as Magnetic Resonance Iimaging (MRI) and computed tomography scans using a computer-aided design. Machine learning and deep learning techniques have gained significance among researchers in medical fields, especially Convolutional Neural Networks (CNN), due to their ability to analyze large amounts of complex image data and perform classification. The objective of this review article is to present an exhaustive study of techniques such as preprocessing, machine learning, and deep learning that have been adopted in the last 15 years and based on it to present a detailed comparative analysis. The challenges encountered by researchers in the past for tumor detection have been discussed along with the future scopes that can be taken by the researchers as the future work. Clinical challenges that are encountered have also been discussed, which are missing in existing review articles.

The existing mission planning models for conventional agile Earth observation satellites assuming stable push-broom imaging cannot be directly adapted to newly developed superagile Earth observation satellites (SA-EOS) with dynamic imaging capabilities. To guarantee the application efficiency of SA-EOS, a novel mission planning modeling method for multipoint target imaging within a single pass (MTI-SP) is proposed. First, a two-step preprocessing method for point targets based on the “nonalong-track” characteristic of SA-EOS converts multiple point targets into imaging strips as an atomic mission for further planning. Then, a triple of decision variables is defined, i.e., the index of the imaging sequence of multiple strips, indices of the push-broom direction of the strips, and normalization coefficients of the imaging time of each endpoint of the strips, on the basis of the one-to-one correspondence between the triple and the imaging time. Third, an optimal MTI-SP mission planning model with two objective functions, namely maximizing the mission coverage benefit (MCB) and minimizing the mission completion time (MCT), is established by considering the imaging time window and attitude transition time. An improved particle swarm optimization algorithm is used for the MTI-SP solution. The results of comparison experiments indicate that the proposed model can achieve the maximum MCB with lower computational consumption. Compared with the two fixed push-broom sequence models and the independent point targets model, the MCT of the proposed model is improved by 47.1%, 24.56%, and 89.45%, respectively. The experiment results show that the proposed modeling method can be further applied in ground operational systems of SA-EOS.

Although significant progress in automatic learning of steganographic cost has been achieved recently, existing methods designed for spatial images are not well applicable to JPEG images which are more common media in daily life. The difficulties of migration mostly lie in the unique and complicated JPEG characteristics caused by 8x8 DCT mode structure. To address the issue, in this paper we extend an existing automatic cost learning scheme to JPEG, where the proposed scheme called JEC-RL (JPEG Embedding Cost with Reinforcement Learning) is explicitly designed to tailor the JPEG DCT structure. It works with the embedding action sampling mechanism under reinforcement learning, where a policy network learns the optimal embedding policies via maximizing the rewards provided by an environment network. The policy network is constructed following a domain-transition design paradigm, where three modules including pixel-level texture complexity evaluation, DCT feature extraction, and mode-wise rearrangement, are proposed. These modules operate in serial, gradually extracting useful features from a decompressed JPEG image and converting them into embedding policies for DCT elements, while considering JPEG characteristics including inter-block and intra-block correlations simultaneously. The environment network is designed in a gradient-oriented way to provide stable reward values by using a wide architecture equipped with a fixed preprocessing layer with 8x8 DCT basis filters. Extensive experiments and ablation studies demonstrate that the proposed method can achieve good security performance for JPEG images against both advanced feature based and modern CNN based steganalyzers.

The performance of stereo vision tasks degrades when haze exists in the input stereo image pair. Independently applying single image dehazing algorithm on left and right images is not optimal. To overcome the problem, we propose an effective framework, called SRDNet, for simultaneously dehazing stereo images. The main idea of SRDNet is to make full use of the stereo information from cross views improving dehazing performance. It does not explicitly employ the disparity estimation and the correlation matrix. SRDNet comprises two parts: a weight-sharing coarse dehazing network (WSCDN) and a guided separated refinement network (GSRN). The WSCDN is utilized to predict a coarse dehazed image pair. Then the GSRN is introduced to predict the residues for different views by extracting the fused information of cross views and separating the features of different views with a guided channel and spatial refinement module. The residues are added to the coarse dehazed pair so as to make refinement and remove the remained haze. Experimental results demonstrate that our proposed SRDNet surpasses previous image dehazing methods by a significant margin both quantitatively and qualitatively. Moreover, our SRDNet could be a preprocessing step of the stereo-based 3D object detection and boosts the 3D detection accuracy in hazy scenes.

BackgroundThe association between brain metabolic change and ischemic stroke has attracted a lot of attention in the research community. 18F-fluorodeoxyglucose (FDG) positron emission tomography (PET) imaging is widely used to measure the metabolism. In experiments, ischemic stroke is usually induced through middle cerebral artery occlusion (MCAO), and quality assessment of this procedure is of vital importance. However, an assessment method based on FDG PET images is still lacking. Herein, we propose an image feature-based protocol to assess the quality of the procedure.MethodsWe performed permanent MCAO to a total of 161 Sprague-Dawley rats. FDG micro-PET images were acquired both before and after the MCAO procedure. Triphenyl tetrazolium chloride (TTC) staining was also conducted to obtain ground truth of the infarct volume. After preprocessing of the PET images, a combination of 3D scale invariant feature transform (SIFT) and support vector machine (SVM) was applied to extract features and train a classifier that can assess the quality of the MCAO procedure.Results106 rats and 212 images were used as training data to construct the classification model. The SVM classifier achieved over 98% accuracy in cross validation. 10 rats with TTC results showing infarction in the ipsilateral brain region served as validation data. Their images were tested by the classifier and all of them were categorized into the correct group. Finally, the remaining 45 rats from a separate experiment were treated as independent test data. The prediction accuracy for these 90 images reached the level of 91%. An online interface was constructed for users to upload their images and obtain the assessment results.ConclusionThis feature-based protocol provides a convenient, accurate and reliable tool to assess the quality of the MCAO procedure in FDG PET study.

Fish classification (FC) is an expansively studied problem in the domains of image segmentation, pattern recognition, and information retrieval. It has been applied in a countless number of domains including target marketing. Meanwhile, governments are obliged to maintain the fish supply and balance between the ecosystem, commercial, agriculture field, marine scientists, and industrial arena of fish including the nutrition and canning factories. The various FC techniques performance is compared relying on the availability of preprocessing and feature extraction methods, the number of extracted features and classification accuracy, the number of fish families/species recognized. This survey also reviewed the use of Databases such as Fish4-Knowledge (F4K), knowledge database, and Global Information System (GIS) on Fishes and other FC databases. The study on preprocessing methods features extraction techniques and classifiers are gathered from recent works to enhance the understanding of the characteristics of preprocessing methods, features extraction techniques, and classifiers to guide future research directions and compensate for current research gaps.

—In graph neural networks (GNNs), pooling operators compute local summaries of input graphs to capture their global properties, and they are fundamental for building deep GNNs that learn hierarchical representations. In this work, we propose the Node Decimation Pooling (NDP), a pooling operator for GNNs that generates coarser graphs while preserving the overall graph topology. During training, the GNN learns new node representations and fits them to a pyramid of coarsened graphs, which is computed offline in a pre-processing stage. NDP consists of three steps. First, a node decimation procedure selects the nodes belonging to one side of the partition identified by a spectral algorithm that approximates the MAXCUT solution. Afterwards, the selected nodes are connected with Kron reduction to form the coarsened graph. Finally, since the resulting graph is very dense, we apply a sparsification procedure that prunes the adjacency matrix of the coarsened graph to reduce the computational cost in the GNN. Notably, we show that it is possible to remove many edges without significantly altering the graph structure. Experimental results show that NDP is more efficient compared to state-of-the-art graph pooling operators while reaching, at the same time, competitive performance on a significant variety of graph classification tasks. submittedVersion

The rapidly increasing incidence of Diabetes Mellitus (DM) has shown that DM is a serious disease that endangered human life in all parts of the world. The late stage of Type-II DM (T2DM) in particular is accompanied by complex complications. Healthcare systems with various data mining algorithms can help the endocrinologist to find whether patients have diabetes in the early detection of T2DM. In the present research, a novel and efficient binary logistic regression (BLR) is proposed founding on feature transformation of XGBoost (XGBoost-BLR) for accurately predicting the specific type of T2DM, and making the model adaptive to more than one dataset. In order to raise the identification ratio, the databases are executed by series of preprocessing procedures which include removing outliers, normalization, and missing value processing. We select features that have a more significant effect on the results by χ 2 test (CST). Then, the selected features are projected into high-dimensional feature space by XGBoost. Finally, the high-dimensional features generated can be modelled by the BLR application. The proposed XGBoost-BLR achieved a 94% and 98% identification rate for diabetes prediction in Pima Indians Diabetes Database (PIDD) and Early-Stage Diabetes Risk Prediction Database (ESDRPD).

Fetching useful information from big medical datasets is a complicated task in the big data age. Various classification algorithms are used in the data mining process to analyze information from the big medical dataset. Nevertheless, these classification algorithms are insufficient to handle big medical data. This work proposes an efficient, ensemble-based classification framework for big medical data to deal with this problem. The proposed work involves initially applying the preprocessing technique to remove noise, missing values, and unwanted features from big medical data. The process selects a subset of classifiers from a pool of classifiers. The selected classifiers are combined to form a hybrid system for efficient classification. The methodology further involves incremental learning from data samples, explaining the predicted outputs, and achieving high classification performance. Java is used for simulation, and the Cleveland Heart Disease big dataset and Diabetes big dataset are used for classification. The experimental result shows that the proposed ensemble algorithm provides an efficient classification compared with existing algorithms based on accuracy, precision, F-measure, recall, and execution time.

Text summarization automatically produces a summary containing important sentences and includes all relevant important information from the original document. One of the main approaches, when viewed from the summary results, are extractive and abstractive. An extractive summary is heading towards maturity and now research has shifted towards abstractive summation and real-time summarization. Although there have been so many achievements in the acquisition of datasets, methods, and techniques published, there are not many papers that can provide a broad picture of the current state of research in this field. This paper provides a broad and systematic review of research in the field of text summarization published from 2008 to 2019. There are 85 journal and conference publications which are the results of the extraction of selected studies for identification and analysis to describe research topics/trends, datasets, preprocessing, features, techniques, methods, evaluations, and problems in this field of research. The results of the analysis provide an in-depth explanation of the topics/trends that are the focus of their research in the field of text summarization; provide references to public datasets, preprocessing and features that have been used; describes the techniques and methods that are often used by researchers as a comparison and means for developing methods. At the end of this paper, several recommendations for opportunities and challenges related to text summarization research are mentioned.

To improve the feature extraction of ship-radiated noise in a complex ocean environment, a novel feature extraction method for ship-radiated noise based on complete ensemble empirical mode decomposition with adaptive selective noise (CEEMDASN) and refined composite multiscale fluctuation-based dispersion entropy (RCMFDE) is proposed. CEEMDASN is proposed in this paper which takes into account the high frequency intermittent components when decomposing the signal. In addition, RCMFDE is also proposed in this paper which refines the preprocessing process of the original signal based on composite multi-scale theory. Firstly, the original signal is decomposed into several intrinsic mode functions (IMFs) by CEEMDASN. Energy distribution ratio (EDR) and average energy distribution ratio (AEDR) of all IMF components are calculated. Then, the IMF with the minimum difference between EDR and AEDR (MEDR) is selected as characteristic IMF. The RCMFDE of characteristic IMF is estimated as the feature vectors of ship-radiated noise. Finally, these feature vectors are sent to self-organizing map (SOM) for classifying and identifying. The proposed method is applied to the feature extraction of ship-radiated noise. The result shows its effectiveness and universality.

Recently, electrocardiogram (ECG) signals have received a high level of attention as a physiological signal in the field of biometrics. It has presented great possibilities for its strength against counterfeit. However, the ECG feature templates are irreplaceable, and a compromised template implies a permanent loss of identity. Therefore, several studies have been introduced biometric template protection techniques such as cancelable techniques to protect the original template in case it is stolen or lost. In this research, a cancelable ECG approach is proposed to protect the ECG feature template for human authentication. In our system, we first employed some image processing techniques for preprocessing the input ECG signals. Then, a deep transfer learning approach is employed to extract the deep ECG features. Later, the proposed cancelable approach based on DNA and amino acid is applied to protect the deep feature templates. Lastly, a Support Vector Machine (SVM) is employed for authentication. Extensive experiments on two commonly used datasets coupled with comprehensive theoretical analysis demonstrate the highest accuracy of the proposed system and the strong resilience of the system to various security and privacy attacks. Results show that the proposed cancelable method meets all requirements of cancelable biometrics such as irreversibility, revocability, and unlinkability.

Spectrum sensing for the Digital Television Terrestrial Multimedia Broadcasting (DTMB) system is considered in this paper. A novel method, which contains a preprocessing method of partition and averaging, a convolutional neural network (CNN) and a threshold chosen scheme, is proposed to solve this problem. In the spectrum sensing process, the received signal is firstly preprocessed by a scheme of partition and averaging. Then it is fed into the well-trained CNN. The posterior probability of the signal being noise is output. Based on the Neyman-Pearson theorem, a threshold can be chosen by the Monte Carlo method. The result of spectrum sensing can be obtained by comparing the posterior probability with the threshold and the complexity of the proposed method is also analyzed. Through computer simulations, the advantage of the proposed method is demonstrated. At a low signal-to-noise ratio (SNR), the proposed method can achieve a satisfactory detection probability. Besides, the proposed method is robust to different SNRs and has good generalization performance under different noise assumptions. When the proposed method is transferred to sense DTMB signal under time dispersive channels, the performance does not deteriorate and even can be better, which shows the robustness of the proposed method to the time dispersive channels.

In this article, for the first time, we propose channel characteristic-based deep neural network (DNN) models for accurate eye-height (EH) and eye-width (EW) estimation of high bandwidth memory (HBM) silicon interposer channels. The proposed models preprocess the design parameters that are highly relevant to the characteristics of the HBM channels. By taking account of the contribution of each design parameter to the eye diagram, the proposed models can accurately estimate the EH and EW of the channels even with a limited number of datasets. For verification, the proposed DNN models were applied to the microstrip and stripline channels of an HBM silicon interposer. Only redistributed layer (RDL) was used to clearly see the effect of reflecting the channel characteristics of the proposed method. We compared the proposed DNN models with various regression methods and a conventional fully connected multilayer DNN model. As a result, the proposed DNN models reduced the EH and EW error rates by 22.7 and 43.9% compared to the other regression methods. In addition, the proposed DNN models not only reduced the error rates by 22.0–28.4% but also reduced the computing cost by 8.0–9.4%, compared to the conventional DNN model. Moreover, we compared the proposed models with various DNN models having other preprocessing structures. By showing 26.7 and 28.8% lower EH and EW error rates than the other DNN models, we validated that the proposed models properly consider the most dominant design factors in preprocessing.

Smooth filtering, a common blurring and denoising operator, has often been utilized postoperatively to diminish the traces left by malicious manipulations. Most of the existing forensic methods only focus on one specific filtering artifact such as median filtering, which is insufficient to reveal the manipulation history of digital images. Unlike traditional convolutional neural network (CNN)-based networks, which normally introduce handcrafted features, including frequency domain features and median filtering residuals, into the preprocessing layer, this paper proposes an end-to-end deep learning model for robust smooth filtering identification. First, a distinctive network structure named the Squeeze-and-Excitation (SE) block is introduced to select discriminative features adaptively and suppress the irrelevant features to the smooth filtering effect. Then, as the network depth increases, multiple inception-residual blocks are stacked to extract discriminative features and reduce the information loss. Finally, different smooth filtering operations can be classified through learning hierarchical features. The experimental results on a composite database show that the proposed model outperforms the state-of-the-art methods, especially in small size and JPEG compression scenarios.

Combining multiple LiDARs enables a robot to maximize its perceptual awareness of environments and obtain sufficient measurements, which is promising for simultaneous localization and mapping (SLAM). This paper proposes a system to achieve robust and simultaneous extrinsic calibration, odometry, and mapping for multiple LiDARs. Our approach starts with measurement preprocessing to extract edge and planar features from raw measurements. After a motion and extrinsic initialization procedure, a sliding window-based multi-LiDAR odometry runs onboard to estimate poses with online calibration refinement and convergence identification. We further develop a mapping algorithm to construct a global map and optimize poses with sufficient features together with a method to model and reduce data uncertainty. We validate our approach's performance with extensive experiments on ten sequences (4.60km total length) for the calibration and SLAM and compare them against the state-of-the-art. We demonstrate that the proposed work is a complete, robust, and extensible system for various multi-LiDAR setups. The source code, datasets, and demonstrations are available at https://ram-lab.com/file/site/m-loam., 20 pages, 22 figures, accepted by IEEE Transcation on Robotics

Electroencephalogram (EEG) signal classification is used in many applications. Typically, this classification is implemented based on methods which consist of two steps. These steps are known as the step of signal preprocessing and the step of the classification. The signal preprocessing step transforms initial signal into classification attributes. According to several studies, this transformation can result in the loss of some useful information and, consequently, the formed classification attributes are uncertain. This information loss can be taken into account if the classification attributes are fuzzy and the fuzzy classifiers are used at the step of classification itself. The transformation of initial EEG signal into fuzzy attributes needs one more procedure at the step of signal preprocessing. This procedure is fuzzification. An approach based on fuzzy classifiers for EEG signal classification is considered in this article. The approach is evaluated based on two classifiers: fuzzy decision tree and fuzzy random Forest. The classification accuracy is 99.5% for fuzzy decision tree and 99.3% for fuzzy random forest. The comparison with similar studies based on non-fuzzy classifiers indicates that fuzzy classifiers are effective tool for EEG signal classification and have best classification accuracy.

Automated brain tumor segmentation of MR image is a very challenging task in a medical point of view. As the nature of the tumor, it can appear anywhere in the brain region with any size, shape, and contrast, that makes the segmentation process more difficult. In order to handle such issues, present work proposes an automated brain tumor segmentation method using rough-fuzzy C-means (RFCM) and shape based topological properties. In rough-fuzzy C-means, overlapping partition is efficiently handled by fuzzy membership and uncertainty in the datasets is resolved by lower and upper bound of the rough set. Fuzzy boundary and crisp lower approximation in RFCM play an effective contribution in brain tumor segmentation on MR images. Initial centroids selection is a major issue in C-means algorithms. Present work has introduced a method for initial centroids selection by which the execution time of RFCM is reduced as compared to random initial centroids. A patch based K-means method is also implemented for skull stripping as a preprocessing step. The proposed method was tested on MRI standard benchmark datasets. Experimental results show that the proposed method has achieved better performance based on statistical volume metrics than previous state-of-the-art algorithms with respect to ground truth (manual segmentation). It is also experimentally noticed that RFCM method achieves most promising results with higher accuracy than HCM (hard C-means) and FCM (fuzzy C-means).

Online condition monitoring and predictive maintenance are crucial for the safe operation of equipments. This paper highlights an unsupervised statistical algorithm based on principal component analysis (PCA) for the predictive maintenance of industrial induced draft (ID) fan. The high vibration issues in ID fans cause the failure of the impellers and, sometimes, the complete breakdown of the fan-motor system. The condition monitoring system of the equipment should be reliable and avoid such a sudden breakdown or faults in the equipment. The proposed technique predicts the fault of the ID fan-motor system, being applicable for other rotating industrial equipment, and also for which the failure data, or historical data, is not available. The major problem in the industry is the monitoring of each and every machinery individually. To avoid this problem, three identical ID fans are monitored together using the proposed technique. This helps in the prediction of the faulty part and also the time left for the complete breakdown of the fan-motor system. This helps in forecasting the maintenance schedule for the equipment before breakdown. From the results, it is observed that the PCA-based technique is a good fit for early fault detection and getting alarmed under fault condition as compared with the conventional methods, including signal trend and fast Fourier transform (FFT) analysis.

Liquid Crystal Display (LCD) defects detection on module-level is increasingly important for flat-panel displays (FPD) industry to increase the production capacity via machine vision technology. However, it is an overwhelmingly challenging issue due to various difficulties. This paper discloses a practical automatic optical inspection (AOI) system consisting of hardware structure and software algorithm to detect module-level defects. The AOI system is the core component to build a distributed integrated inspection system with the help of the Internet of Things (IoT). Starting from the analysis of the challenges encountered in module-level defects inspection, a delicate photograph scheme is proposed to reveal different kinds of defects. In order to robustly work on the module-level defects detection with complex situations, a novel framework based on YOLOV3 detection unit is proposed in this paper, including the preprocessing module, detection module, defects definition module and interferences elimination module. To the best of our knowledge, this is the first work that designs a practical automatic optical inspection system for module-level defects detection. In order to demonstrate the effectiveness of the proposed method, extensive experiments have been conducted on the manufacturing lines. Evaluation of the detection performance of the AOI system in comparison with a manual scheme indicates that the proposed system is practical for module-level defects detection. Currently, the proposed system has been deployed in a real-world LCD manufacturing line from a major player in the world.

Vehicle detection in remote sensing images remains a challenge because most vehicles are small and cover only a relatively small area due to the low ground sample distance. Although image super-resolution can improve small object detection performance as a preprocessing step, methods for improving the quality of the entire image tend to focus on the majority backgrounds that are not important for detection and involve high computational cost. Inspired by the promising feature-level super-resolution method, in this letter, we propose a novel anchor-free vehicle detection network for small vehicle detection in remote sensing images. Specifically, a target-guided feature super-resolution network is proposed to enhance the features of the potential target. Besides, we propose a novel feature fusion module to improve the feature representation of shallow layers, which accounts for small object detection. Extensive experiments on three public remote sensing detection datasets [cars overhead with context (COWC), Vehicle Detection in Aerial Imagery (VEDAI), and UCAS-AOD] amply demonstrate that our method can achieve significant performance with a mean average precision of 0.933, 0.756, and 0.961, respectively.

Internet of Things (IoT) realizes the interconnection of heterogeneous devices by the technology of wireless and mobile communication. The data of target regions are collected by widely distributed sensing devices and transmitted to the processing center for aggregation and analysis as the basis of IoT. The quality of IoT services usually depends on the accuracy and integrity of data. However, due to the adverse environment or device defects, the collected data will be anomalous. Therefore, the effective method of anomaly detection is the crucial issue for guaranteeing service quality. Deep learning is one of the most concerned technology in recent years which realizes automatic feature extraction from raw data. In this article, the integrated model of the convolutional neural network (CNN) and recurrent autoencoder is proposed for anomaly detection. Simple combination of CNN and autoencoder cannot improve classification performance, especially, for time series. Therefore, we utilize the two-stage sliding window in data preprocessing to learn better representations. Based on the characteristics of the Yahoo Webscope S5 dataset, raw time series with anomalous points are extended to fixed-length sequences with normal or anomaly label via the first-stage sliding window. Then, each sequence is transformed into continuous time-dependent subsequences by another smaller sliding window. The preprocessing of the two-stage sliding window can be considered as low-level temporal feature extraction, and we empirically prove that the preprocessing of the two-stage sliding window will be useful for high-level feature extraction in the integrated model. After data preprocessing, spatial and temporal features are extracted in CNN and recurrent autoencoder for the classification in fully connected networks. Empiric results show that the proposed model has better performances on multiple classification metrics and achieves preferable effect on anomaly detection.