Avoid Information Loss Research Articles

Multi-scale wavelet network algorithm for pediatric echocardiographic segmentation via hierarchical feature guided fusion

The automatic segmentation of critical anatomical structures in pediatric echocardiography is the essential steps for early diagnosis and treatment of congenital heart disease. However, current segmentation algorithms rarely extract the information based on effective feature enhancement algorithms. Simultaneously, the algorithms are susceptible to image quality and the lack of detail information. To solve this, we propose a multi-scale wavelet network (MS-Net) combined with a bidirectional feature fusion (BFF-Net) and a wavelet-Unet (W-Unet) for end-to-end pediatric echocardiographic segmentation. In MS-Net, the entire network uses the discrete wavelet transform (DWT) instead of the sampling operation to reduce the impact of image noise while avoiding information loss. The algorithm enhances the edge information by designing the edge attention module (EAM) in the BFF-Net branch and fuses the context and detail information via the bidirectional feature fusion. Secondly, this algorithm uses W-Unet to obtain the detail features of high-resolution images by network depth reduction and propagation method, which supplements the features extracted by the BFF-Net branch. Finally, the hierarchical features of BFF-Net and W-Unet are fused and updated by guided filtering (GF) to obtain the final segmentation prediction. Using 127 pediatric echocardiographic cases of self-selection as the experimental dataset, the left atrium and left ventricle of the echocardiogram were segmented. The Dice coefficient values of 0.9532 and 0.9155, the pixel accuracy of 0.9914, and the specificity values of 0.9975 and 0.9984 were obtained. It was thus verifying its potential and effectiveness as a clinical auxiliary tool.

Model Predictive Control of Nonlinear System Based on GA-RBP Neural Network and Improved Gradient Descent Method

A model predictive control (MPC) method based on recursive backpropagation (RBP) neural network and genetic algorithm (GA) is proposed for a class of nonlinear systems with time delays and uncertainties. In the offline modeling stage, a multistep-ahead predictor with GA-RBP neural network is designed, where GA-BP neural network is used as a one-step prediction model and GA is employed to train the initial weights and bias of the BP neural network. The incorporation of GA into RBP can reduce the possibility of the BP neural network falling into a local optimum instead of reaching global optimization. In the online optimizing stage, a multistep-ahead GA-RBP neural network predictor and an improved gradient descent method (IGDM) are proposed to efficiently solve the online optimization problem of nonlinear MPC by minimizing a modified quadratic criterion. The designed MPC strategy can avoid information loss while linearizing the controlled system and computing the Hessian matrix and its inverse matrix. Experimental results show that the proposed approach can reduce the computational burden and improve the performance of MPC (i.e., the maximum overshoots, calculation time, rise time, and RMSE tracking error value) for the solution of nonlinear controlled systems.

NAS-Guided Lightweight Multiscale Attention Fusion Network for Hyperspectral Image Classification

Deep learning (DL) has become a hot topic in the research field of hyperspectral image (HSI) classification. However, with increasing depth and size of deep learning methods, its application in mobile and embedded vision applications has brought great challenges. In this article, we address a network architecture search (NAS)-guided lightweight spectral–spatial attention feature fusion network (LMAFN) for HSI classification. The overall architecture of the proposed network is guided by several conclusions of NAS, which achieves fewer parameters and lower computation cost with deeper network structure by exploiting multiscale Ghost grouped with efficient channel attention (ECA) module for adaptively adjusting the weights of different channels. It helps fully extract spectral–spatial discriminant features to avoid information loss of the dimension reduction operation. Specifically, a multilayer feature fusion method is proposed to extract the fusion information of the spectral–spatial features of each layer by considering complementary information of different hierarchical structures. Therefore, high-lever spectral–spatial attributes are gradually exploited along with the increase in layers and the fusion of layers. The experimental verification on three real HSI data sets demonstrates that the proposed framework presents more satisfying classification performance and efficiency with deeper network structure and lower parameter size.

A Novel Auction-Based Optimization Algorithm and Its Application in Rough Set Feature Selection

The selection of features from data, as one of the most important tasks in data mining, strongly affects the accuracy of classification. The removal of irrelevant and redundant features from data while simultaneously avoiding information loss is the main objective of feature selection. Feature selection is possible using rough set theory and meta-heuristic algorithms. In this paper, a novel auction-based optimization algorithm (ABOA) is proposed to contribute to generating an effective algorithm with a good trade-off between exploration and exploitation. This new algorithm simulates the auction sale process, where bidders offer higher/lower amounts to outbid each other. Auctions are categorized into ascending auctions and descending auctions and thus respectively represent maximization and minimization problems in ABOA. In the first step of the ABOA after initialization, a predefined number of bidders is selected and an auction is performed between them. The winner is selected and another auction is performed between the winner and a predefined number of the winner’s neighbors. The winner of this round of auction is added to the winner list. This process is iterated until a predefined number of winners is found. Finally, one more auction is performed between all the winners on the winner list and the winner of that auction becomes the final winner. The algorithm with different parameter setting scenarios is tested on 25 benchmark test functions. The algorithm with the best results is then used to perform feature selection on 18 UCI datasets. The feature selection and classification accuracy results are compared with state-of-the-art results. The statistical analysis of the results proves the ability of the algorithm to solve optimization problems.

Supplier selection based on multi-stakeholder Best- Worst Method

Supplier selection and supplier segmentation are two interdependent crucial steps of effective supplier relationship management. Supplier selection is formulated as a multi-criteria decision-making problem, in which evaluation criteria play a critical role in a realistic assessment of the suppliers. The evaluation criteria can be used to link supplier selection and segmentation steps to have a more efficient and effective supplier relationship management. A multi-stakeholder Best-Worst Method (BWM) is used for supplier selection. The proposed method is a proper tool to handle the optimal weights of the BWM, which are in the form of intervals. The proposed approach enables decision-makers to avoid information loss within their decision-making procedure resulting in more realistic decisions. A real-world case study is used to illustrate the proposed decision-making framework. The final results show the best suppliers based on both capabilities and willingness dimensions which can be selected by the case company.

Clinical Target Volume Auto-Segmentation of Esophageal Cancer for Radiotherapy After Radical Surgery Based on Deep Learning.

Radiotherapy plays an important role in controlling the local recurrence of esophageal cancer after radical surgery. Segmentation of the clinical target volume is a key step in radiotherapy treatment planning, but it is time-consuming and operator-dependent. This paper introduces a deep dilated convolutional U-network to achieve fast and accurate clinical target volume auto-segmentation of esophageal cancer after radical surgery. The deep dilated convolutional U-network, which integrates the advantages of dilated convolution and the U-network, is an end-to-end architecture that enables rapid training and testing. A dilated convolution module for extracting multiscale context features containing the original information on fine texture and boundaries is integrated into the U-network architecture to avoid information loss due to down-sampling and improve the segmentation accuracy. In addition, batch normalization is added to the deep dilated convolutional U-network for fast and stable convergence. In the present study, the training and validation loss tended to be stable after 40 training epochs. This deep dilated convolutional U-network model was able to segment the clinical target volume with an overall mean Dice similarity coefficient of 86.7% and a respective 95% Hausdorff distance of 37.4 mm, indicating reasonable volume overlap of the auto-segmented and manual contours. The mean Cohen kappa coefficient was 0.863, indicating that the deep dilated convolutional U-network was robust. Comparisons with the U-network and attention U-network showed that the overall performance of the deep dilated convolutional U-network was best for the Dice similarity coefficient, 95% Hausdorff distance, and Cohen kappa coefficient. The test time for segmentation of the clinical target volume was approximately 25 seconds per patient. This deep dilated convolutional U-network could be applied in the clinical setting to save time in delineation and improve the consistency of contouring.

Retracted: Analysis and Research of Artificial Intelligence Algorithms in GPS Data

With the development of GPS technology, it has been gradually applied in engineering, science and technology and daily life. The requirement of positioning accuracy in these fields is also increasing. Therefore, this paper studies GPS coordinate conversion and positioning technology from the aspect of artificial intelligence algorithm to improve positioning accuracy. Aiming at the problem of coordinate transformation, a method of GPS coordinate transformation based on convolution neural network model is proposed. Firstly, the input GPS raw data is converted into unstructured images, then the model is used to learn the features of the data, and finally the converted coordinate data is output. Aiming at the problem that the traditional vehicle GPS navigation system is easy to be interfered by the external environment and is not conducive to the system tracking and positioning, AdaBoost algorithm is introduced into the navigation system to better solve the problem. GPS lock. The strong classifier can be obtained by iterative practice, and the filter value can be measured accurately during GPS interference. At the same time, the combination of Adaboost method and BP algorithm can help the filter process information and avoid information loss caused by GPS interference. In this way, not only the integrity of information can be guaranteed, but also the stability and accuracy of the system can be guaranteed. The results show that the coordinate transformation method based on convolution neural network has better conversion accuracy than the traditional method, and the navigation method based on Adaboost algorithm can improve the navigation accuracy and optimize the system performance.

A Two-Branch Network Combined With Robust Principal Component Analysis for Hyperspectral Image Classification

Noise in hyperspectral images (HSIs) may degrade the HSI classification result. Robust principal component analysis (RPCA) is an excellent method to obtain low-rank (LR) representation of data and is widely used in image denoising and also in HSI classification. However, data are drawn as a union from multiple subspaces in HSIs, so LR subspace estimation (LRSE) is necessary when using RPCA, which is complicated and time-consuming. To solve this problem, this letter proposes a novel LR-based method for HSI classification called two-branch network combined with RPCA, which combines RPCA with a neural network. Specifically, both the LR component and the sparse component are preserved and used for feature extraction in two independent convolutional branches. This way, we can avoid information loss without using accurate LRSE. A concatenate operation and a pointwise convolution are then adopted to realize the feature fusion. Finally, fused features are constructed to indicate the ground truth of each pixel in the classification process. The proposed method constructs a convenient model for HSI classification by discarding the LRSE and combining denoising, feature extraction, feature fusion, and classification into an end-to-end network. The experimental results on three data sets demonstrate that the proposed method outperforms many state-of-the-art methods including ones based on LR representation and ones based on deep learning. In addition, it maintains good classification performance for the cases of small samples and class imbalance.

Imputing missing indoor air quality data via variational convolutional autoencoders: Implications for ventilation management of subway metro systems

Missing data represents a common problem in environmental and building-related processes, especially in the indoor air quality (IAQ) system of subway stations, where the collected information leads to actions in ventilation management. For these reasons, imputation approaches have been used to avoid information loss due to downsampling or sensor malfunction. This paper introduces an imputation approach for IAQ data via variational autoencoders (VAE) coupled with convolutional layers (VAE-CNN). Two scenarios were introduced: first, the IAQ dataset was corrupted by removing data intervals at different missing rates (i.e., 20%, 50%, and 80%), and second, a point-to-point removal of three sensors was conducted. The performance of the proposed method was compared with different techniques, showing that the VAE-CNN was superior to other methods even for massive amounts of missing data. Finally, the effects of missing and imputed data on the IAQ system in the D-subway station were evaluated in terms of ventilation energy demand, CO2 emissions, and IAQ level. The IAQ management with the imputed data could decrease by approximately 20% of CO2 emissions by reducing the energy demand, while the IAQ level increased by 3% in another scenario.

Mixed data sampling expectile regression with applications to measuring financial risk

To avoid information loss or measurement error in traditional methods dealing with mixed frequency data, we develop a novel mixed data sampling expectile regression (MIDAS-ER) model to measure financial risk. We construct the MIDAS-ER model by introducing a MIDAS structure into expectile regressions. This enables us to perform an expectile regression on raw mixed frequency data directly. We apply the proposed MIDAS-ER model to estimate two popular financial risk measures, namely, Value at Risk and Expected Shortfall, with both simulated data and four stock indices, and compare the model's performance with those of several popular models. The outstanding performance of our model demonstrates that high-frequency information helps to improve the accuracy of risk measurement. In addition, the numerical results also imply that our model can be a significant tool for risk-averse investors to control risk losses and for financial institutions to implement robust risk management.

A novel ResNet-based model structure and its applications in machine health monitoring

Machine health monitoring has become increasingly important in modern manufacturers because of its ability to reduce downtime of the machine and cut down the production cost. Enormous signals acquired from machinery are capable of reflecting current working conditions by in-depth analysis with various data-driven methods. Hand-crafted feature extraction and representation from the traditional methods are essential but daunting tasks, and these methods may not be suitable for these massive data. Compared with traditional methods, deep learning ones are able to extract the best feature combination during model training without any artificial intervention, which makes it easier, more efficient, and more effective to monitor machine health, but the training cost and training time hamper its application. The short-time Fourier transform is adopted as the data preprocessing method to cut down the training cost and boost the training procedure. Inspired by the great achievements of ResNet, the new optimized model based on ResNet has been proposed with layer-by-layer dimension reduction of the feature maps. The proposed model is also able to avoid information loss in the conventional pooling layer. All the potential candidate model blocks are introduced and compared, and the best one is selected as the final one. Repeated model block layers are adapted for the best feature combinations, followed by a two-layer full connection layer for the final targets. The proposed method is validated by conducting experiments on bearing fault diagnosis and tool wear prediction dataset. The final results show that the proposed model achieves the best accuracy rate in the classification task and the lowest root mean squared error in the prediction task.

Limited interval-valued probabilistic linguistic term sets in evaluating airline service quality

The probabilistic linguistic term set (PLTS) is a useful tool and has been widely applied to deal with uncertain information for decision-making problems. It allows the experts to evaluate the alternatives by linguistic terms with corresponding probability information. In practice, the experts are difficult to provide complete probability information because of the complexity of the decision-making environment and limited cognition of the experts. In such cases, we need to normalize the probability information. To avoid information loss in the normalization process of PLTSs, this paper proposes the concept called limited interval-valued probabilistic linguistic term sets (-IVPLTSs) by introducing the membership degree. First, we present the concept of -IVPLTSs, and provide the basic operation laws and aggregation operators for -IVPLTSs. Then, the membership degree is determined by the deviation degree based on a programming model. Furthermore, the extended possibility degree and the PROMETHEE II method under the limited interval-valued probabilistic linguistic environment are given, based on which, the whole multi-criteria group decision making (MCGDM) process with -IVPLTSs is presented. Finally, the proposed method is applied to a case about evaluation of airline service quality. Discussions and analyses about the results are further conducted to verify the rationality of our approach.

Evaluating the impact of a coordinated checkpointing in distributed data streams processing systems using discrete event simulation

Data Streams Processing systems process continuous flows of data under Quality of Service requirements. Data streams often contain critical information which requires real-time processing. To guarantee systems' dependability and avoid information loss, one must use a fault-tolerance strategy. However, there are several strategies available, and the proper evaluation of which mechanism is better for each system architecture is challenging, especially in large-scale distributed systems. In this paper, we propose a discrete simulation model for investigating the impacts of the Coordinated Checkpoint fault tolerance strategy imposes on Data Stream Processing Systems. Results show that this strategy critically affects stream processing in failure-prone situations due to an increase in latency up to 120% and information loss, reaching 95% of the processing window in the worst case.

A New Distance Metric Exploiting Heterogeneous Interattribute Relationship for Ordinal-and-Nominal-Attribute Data Clustering.

Ordinal attribute has all the common characteristics of a nominal one but it differs from the nominal one by having naturally ordered possible values (also called categories interchangeably). In clustering analysis tasks, categorical data composed of both ordinal and nominal attributes (also called mixed-categorical data interchangeably) are common. Under this circumstance, existing distance and similarity measures suffer from at least one of the following two drawbacks: 1) directly treat ordinal attributes as nominal ones, and thus ignore the order information from them and 2) suppose all the attributes are independent of each other, measure the distance between two categories from a target attribute without considering the valuable information provided by the other attributes that correlate with the target one. These two drawbacks may twist the natural distances of attributes and further lead to unsatisfactory clustering results. This article, therefore, presents an entropy-based distance metric that quantifies the distance between categories by exploiting the information provided by different attributes that correlate with the target one. It also preserves the order relationship among ordinal categories during the distance measurement. Since attributes are usually correlated in different degrees, we also define the interdependence between different types of attributes to weight their contributions in forming distances. The proposed metric overcomes the two above-mentioned drawbacks for mixed-categorical data clustering. More important, it conceptually unifies the distances of ordinal and nominal attributes to avoid information loss during clustering. Moreover, it is parameter free, and will not bring extra computational cost compared to the existing state-of-the-art counterparts. Extensive experiments show the superiority of the proposed distance metric.

Validation of Minimal Number of Force Sensitive Resistors to Predict Risk of Falling During a Timed Up and Go Test

Several studies use force sensitive resistors (FSR) to compute gait and balance parameters related to falls without investigating the number of sensor units required to produce useful information. We propose a model with minimal sensors for an instrumented insole by investigating and optimizing the location and variety of sensors required to efficiently detect people at risk of falling. Datasets previously recorded on twelve Parkinson’s disease (PD) participants (67.7 ± 10.07 years), nine healthy elderly (66.8 ± 8.0 years) and ten young healthy adults (28.27 ± 3.74 years) were used in this study. We compared the datasets obtained from the use of four FSRs with those of three, two, one and no FSR; each set was combined with an inertial measurement unit (IMU). During the walking activity, the risk of falling scores from four FSRs and IMU (acceleration in y-axis only) were not significantly different compared with two FSRs and IMU (p > 0.05), whereas significant difference was found for three FSRs and IMU and one FSR and IMU (p 0.05). We concluded that it is feasible to estimate the risk index after reducing the number of sensing units from four to two FSRs during walking test and from four to three FSRs during sit-to-stand and stand-to-sit tests. The FSRs should be placed at strategic positions to avoid information loss.

Building safety monitoring based on extreme gradient boosting in distributed optical fiber sensing

Abstract Aiming at the problem of the safety monitoring of the surrounding environment of the glass window on the building, a scheme of distributed optical fiber acoustic sensors based on the phase sensitive time-domain reflection is proposed. Firstly, Wigner bispectrum is used to analyze the time–frequency energy distribution of vibration signal, then the axially integrated bispectrum algorithm is used to extract the time–frequency characteristic components, and the method of sub-band division is used to avoid information loss and decrease the cost of computing. The eigenvectors of different vibration events are recognized by extreme gradient boosting tree algorithm. Based on the diversity of acoustic signal types in the environment, there are eight kinds of events including wind blowing, knocking, window opening, watering, hammering, dog barking, aircraft sound and no disturbance to be selected for vibration experiments. The experimental results show that the scheme can get better recognition effect, and the average recognition rate of eight kinds of events reaches 93.3%.

Anchor-Free Single Stage Detector in Remote Sensing Images Based on Multiscale Dense Path Aggregation Feature Pyramid Network

Object detection has always been a challenging task in the field of computer vision due to complex background, large scale variation and many small objects, which are especially pronounced for remote sensing imagery. In recent years, object detection in remote sensing with the development of deep learning has also made great breakthroughs. At present, almost all state-of-the-art object detectors rely on pre-defined anchor boxes for remote sensing imagery. However, too many anchor boxes will introduce a large number of hyper-parameters, which not only increase the memory footprint, but also increase the computational redundancy of the detection model. In contrast, we propose an anchor-free single-stage detector for remote sensing imagery object detection, avoiding a large number of hyper-parameters related to the anchor box, which usually affect the performance of the detection model. Specially, considering the large-scale differences in the objects and the characteristics of small objects in remote sensing imagery, we design a dense path aggregation feature pyramid network (DPAFPN), which can make full use of the high-level semantic information and low-level location information in remote sensing imagery, and to a certain extent, avoid information loss during shallow feature transfer. In our experiments, extensive experiments on two public remote sensing datasets DOTA, NWPU VHR-10 were conducted. The experimental results demonstrate that our detector has good performance and is meaningful for object detection in remote sensing imagery.

Asymmetric Deeply Fused Network for Detecting Salient Objects in RGB-D Images

Most RGB-D salient object detection (SOD) models use the same network to process RGB images and their corresponding depth maps. Subsequently, these models perform direct concatenation and summation at deep or shallow layers. However, these models ignore the complementarity of multi-level features extracted from RGB images and depth maps. This paper presents an asymmetric deeply fused network (ADFNet) for RGB-D SOD. Two different backbone networks, i.e., ResNet-50 and VGG-16, are utilized to process RGB images and related depth maps. We use an aggregation decoder and adaptive attention transformer module (AATM) to avoid information loss in the decoding process. Additionally, we use an attention early fusion module (AEFM) and deep fusion module (DFM) to deal with the deep features in various complex situations. Experiments validate the effectiveness of the proposed ADFNet, which outperforms thirteen recent RGB-D SOD models in the analysis of five public RGB-D SOD datasets.

Method of rock mass quality classification in ancient underground engineering based on three-parameter interval grey number

Rock mass quality classification is an important basis for long-term protection and tourism development of ancient underground engineering, however, the existing rock mass quality classification methods are difficult to apply to the special project of ancient underground engineering. Based on the in-depth analysis of the characteristics of ancient underground engineering and the influencing factors of rock mass quality, an evaluation index system for rock mass quality classification of ancient underground engineering is established by taking the long-term tensile strength of rock and the weathering velocity of rock mass as indicators. Secondly, for the evaluation index value has the character of bounded-but-unknown uncertainty, and in order to avoid information loss in interval numbers, the three-parameter interval grey number theory is introduced, based on the concept of three-parameter interval grey number distance measure, a rock mass quality classification calculation model for ancient underground engineering is established. Then, because of the complexity of ancient underground engineering, the idea of group decision-making is introduced, and based on the variation coefficient method, the weight calculation method of evaluation index is established to improve the reliability of weight calculation. Finally, a rock mass quality classification method for ancient underground engineering based on three-parameter interval grey number is established. The method is used to analysis the practical engineering, it shows that the method proposed here is feasible and reasonable.

Study on the Intensity and Coherence Information of High-Resolution ALOS-2 SAR Images for Rapid Massive Landslide Mapping at a Pixel Level

A rapid mapping of landslides following a disaster is important for coordinating emergency response and limiting rescue delays. A synthetic aperture radar (SAR) can provide a solution even in harsh weather and at night, due to its independence of weather and light, quick response, no contact and broad coverage. This study aimed to conduct a comprehensive exploration on the intensity and coherence information of three Advanced Land Observing Satellite-2 (ALOS-2) SAR images, for rapid massive landslide mapping in a pixel level, in order to provide a reference for future applications. Applied data were two pre-event and one post-event high-resolution ALOS-2 products. Studied area was in the east of Iburi, Hokkaido, Japan, where massive shallow landslides were triggered in the 2018 Hokkaido Eastern Iburi Earthquake. Potential parameters, including intensity difference (d), co-event correlation coefficient (r), correlation coefficient difference ( ∆ r ), co-event coherence ( γ ), and coherence difference ( ∆ γ ), were first selected and calculated based on a radar reflection mechanism, to facilitate rapid detection. Qualitative observation was then performed by overlapping ground truth landslides to calculated parameter images. Based on qualitative observation, an absolute value of d ( d a b s 1 ) was applied to facility analyses, and a new parameter ( d a b s 2 ) was proposed to avoid information loss in the calculation. After that, quantitative analyses of the six parameters ( d a b s 1 , d a b s 2 , r, ∆ r , γ and ∆ γ ) were performed by receiver operating characteristic. d a b s 2 and ∆ r were found to be favorable parameters, which had the highest AUC values of 0.82 and 0.75, and correctly classified 69.36% and 64.57% landslide and non-landslide pixels by appropriate thresholds. Finally, a discriminant function was developed, combining three relatively favorable parameters ( d a b s 2 , ∆ r , and ∆ γ ) with one in each type, and achieved an overall accuracy of 74.31% for landslide mapping.