Search

Power line inspection is an important means to eliminate hidden dangers of power lines. It is a difficult research problem how to solve the low accuracy of power line inspection based on deep neural network (DNN) due to the problems of multi‐view‐shape, small‐size object. In this paper, an automatic detection model based on Feature visual clustering network (FVCNet) for power line inspection is established. First, an unsupervised clustering method for power line inspection is proposed, and applied to construct a detection model which can recognize multi‐view‐shape objects and enhanced object features. Then, the bilinear interpolation method is used to Feature enhancement method, and the enhanced high‐level semantics and low‐level semantics are fused to solve the problems of small object size and single sample. In this paper, FVCNet is applied to the MS‐COCO 2017 data set and self‐made power line inspection data set, and the test accuracy is increased to 61.2% and 82.0%, respectively. Compared with other models, especially for the categories that are greatly affected by multi‐view‐shape, the test accuracy has been improved significantly. [ABSTRACT FROM AUTHOR]

This paper trains a multilayered neural network (MLNN) for solving fractional delay differential equations (FDDEs), including nonlinear and singular types. The proposed methodology involves replacing the unknown functions in the equations with a truncated power series expansion. Subsequently, a collection of algebraic equations is solved utilizing an iterative minimization technique that leverages the capabilities of the MLNN architecture. The outcomes demonstrate that the MLNN architecture provides the required accuracy and strong stability compared to several numerical methods. [ABSTRACT FROM AUTHOR]

The driver authority decision for driver‐initiated takeover is closely related to vehicle driving safety. In this paper, a lateral driver‐automation driver authority decision method considering the safety of the intended functionality is proposed to improve vehicle driving safety. Based on systems‐theoretic processes analysis, functional safety analysis of driver initiative takeover is carried out to clarify the safety of the intended functionality issues caused by the unreasonable decision. Then, a method for defining the safe driving area is developed to assess the safety level of the driving area around the ego vehicle. Taking the driver‐vehicle state parameters and vehicle–vehicle state parameters as input, a deep neural network model is constructed to determine the intention of the driver to take over the vehicle. The lateral driver‐automation driver authority decision method is designed to make the driver‐machine take‐over decisions, which considers the safety of the intended functionality. The effectiveness of the proposed method is evaluated via numerical simulation and hardware‐in‐the‐loop experiments. The results show that the designed method not only improves the control ability of the driver to the maximum extent but also integrates the perception ability of the driver into the vehicle control system to improve further vehicle driving safety. [ABSTRACT FROM AUTHOR]

Many test coverage metrics have been proposed to measure the deep neural network (DNN) testing effectiveness, including structural coverage and nonstructural coverage. These test coverage metrics are proposed based on the fundamental assumption: They are correlated with test effectiveness. However, the fundamental assumption is still not validated sufficiently and reasonably, which brings question on the usefulness of DNN test coverage. This paper conducted a revisiting study on the existing DNN test coverage from the test effectiveness perspective, to effectively validate the fundamental assumption. Here, we carefully considered the diversity of subjects, three test effectiveness criteria, and both typical and state‐of‐the‐art test coverage metrics. Different from all the existing studies that deliver negative conclusions on the usefulness of existing DNN test coverage, we identified some positive conclusions on their usefulness from the test effectiveness perspective. In particular, we found the complementary relationship between structural and nonstructural coverage and identified the practical usage scenarios and promising research directions for these existing test coverage metrics. [ABSTRACT FROM AUTHOR]

In this paper, the geometry factors of mode I and mode II (YI and YII) and the normalized T‐stress (T*) of the centrally cracked Brazilian disk specimen are predicted using a deep learning approach. Three deep neural networks are developed to model the relationship between the crack angle (α) and the ratio of half the crack length to the radius (a/R) as input variables and each of YI, YII, and T* as output variables. Three independent databases consisting of 174, 174, and 117 data points are prepared for YI, YII, and T*, respectively, from the previous works to train, generalize, and validate the deep neural networks. Finally, sensitivity analysis of YI, YII, and T* to α and a/R is conducted, and mathematical models are obtained using the partial dependence plots, which can be used in the optimization of stress intensity factors and T‐stress of the cracked Brazilian disk specimens with different dimensions. Highlights: Three analytical databases are prepared for YI, YII, and T*.Three independent DNN models are developed for predicting YI, YII, and T* of the BD specimen.The sensitivity of YI, YII, and T* to the α and a/R are evaluated using PDP analysis.The mathematical relationship between α and a/R, and YI, YII, and T* are obtained. [ABSTRACT FROM AUTHOR]

Price impact is the adverse change of the asset price against trader's action. As a crucial part of the indirect trading cost, price impact has attracted increasing attention in both econometric and data science literature. In this paper, we draw upon both strands of the literature and develop a deep neural network enhanced recursive (DeRecv) model to estimate temporary and permanent price impact of an order or trade. The temporary price impact is calculated as the sum of the expected immediate impact at each time point after taking action in an ad hoc market condition. The permanent price impact is defined as a new permanent level at which the information of the incoming order is entirely absorbed by the market. Through the experimental evaluation based on data from 10 stocks at NASDAQ and Shanghai Stock Exchange, we show that the proposed DeRecv model is better than the reinforcement learning model and the traditional vector autoregressive model. [ABSTRACT FROM AUTHOR]

The syndrome called COVID‐19 which was firstly spread in Wuhan, China has already been declared a globally "Pandemic." To stymie the further spread of the virus at an early stage, detection needs to be done. Artificial Intelligence‐based deep learning models have gained much popularity in the detection of many diseases within the confines of biomedical sciences. In this paper, a deep neural network‐based "LiteCovidNet" model is proposed that detects COVID‐19 cases as the binary class (COVID‐19, Normal) and the multi‐class (COVID‐19, Normal, Pneumonia) bifurcated based on chest X‐ray images of the infected persons. An accuracy of 100% and 98.82% is achieved for binary and multi‐class classification respectively which is competitive performance as compared to the other recent related studies. Hence, our methodology can be used by health professionals to validate the detection of COVID‐19 infected patients at an early stage with convenient cost and better accuracy. [ABSTRACT FROM AUTHOR]

Purpose: Fetal head circumference (HC) is an important biometric parameter that can be used to assess fetal development in obstetric clinical practice. Most of the existing methods use deep neural network to accomplish the task of automatic fetal HC measurement from two‐dimensional ultrasound images, and some of them achieved relatively high prediction accuracy. However, few of these methods focused on optimizing model efficiency performance. Our purpose is to develop a more efficient approach for this task, which could help doctors measure HC faster and would be more suitable for deployment on devices with scarce computing resources. Methods: In this paper, we present a very lightweight deep convolutional neural network to achieve automatic fetal head segmentation from ultrasound images. By using sequential prediction network architecture, the proposed model could perform much faster inference while maintaining a high prediction accuracy. In addition, we used depthwise separable convolution to replace part of the standard convolution in the network and shrunk the input image to further improve model efficiency. After getting fetal head segmentation results, post‐processing, including morphological processing and least‐squares ellipse fitting, was applied to obtain the fetal HC. All experiments in this work were performed on a public dataset, HC18, with 999 fetal ultrasound images for training and 335 for testing. The dataset is publicly available on https://hc18.grand‐challenge.org/ and the code for our method is also publicly available on https://github.com/ApeMocker/CSM‐for‐fetal‐HC‐measurement. Results: Our model has only 0.13 million [M] parameters and can achieve an inference speed of 28 [ms] per frame on a CPU and 0.194 [ms] per frame on a GPU, which far exceeds all existing deep learning‐based models as far as we know. Experimental results showed that the method achieved a mean absolute difference of 1.97(± 1.89) [mm] and a Dice similarity coefficient of 97.61(± 1.72) [%] on HC18 test set, which were comparable to the state of the art. Conclusion: We presented a very lightweight deep learning‐based model to realize fast and accurate fetal head segmentation from two‐dimensional ultrasound image, which is then used for calculating the fetal HC. The proposed method could help obstetricians measure the fetal HC more efficiently with high accuracy, and has the potential to be applied to the situations where computing resources are relatively scarce. [ABSTRACT FROM AUTHOR]

Inverse modeling of microwave components plays an important role in microwave design and diagnosis or tuning. Since the analytical function or formula of the inverse input‐output relationship does not exist and is difficult to obtain, artificial neural network (ANN) becomes an efficient tool to develop inverse models for microwave components. This paper provides an overview of recent advances in neural network‐based inverse modeling techniques for microwave applications. We review two different shallow neural network‐based inverse modeling techniques, including the comprehensive neural network inverse modeling methodology and the multivalued neural network inverse modeling technique. Both techniques address the problem of nonuniqueness in inverse modeling. We also provide an overview of recently developed hybrid deep neural network modeling technique and the application to inverse modeling. For the inverse modeling problem with high‐dimensional inputs, the relationship between the inputs and the outputs of the inverse model will become more complicated and the inverse modeling problem will become harder. The deep neural network becomes a practical choice. The hybrid deep neural network structure is presented. The recently proposed activation function, specifically for microwave application, and a three‐stage deep learning algorithm for training the hybrid deep neural network are reviewed. [ABSTRACT FROM AUTHOR]

Visual interpretation of chest X‐rays (CXRs) is tedious and prone to error. Significant amount of time is spent by the radiologist in differentiating normal from abnormal CXRs and in identifying the location and type of abnormalities. An assistance tool for automatically classifying normal and different types of abnormal CXRs can facilitate the diagnosis and potentially save time costs. In this paper, a novel hybrid model having concatenation of Visual Geometry Group (VGG19) network and Entropy features as a modified deep convolutional neural network (DCNN) architecture, called VEntNet, is proposed for the automated multi‐class categorization of CXR images into normal, coronavirus disease (COVID), tuberculosis (TB), viral pneumonia, and bacterial pneumonia. The VEntNet model implemented consists of deep features extraction from convolutional layers of VGG19 network which are then concatenated with hand‐crafted entropy features extracted from CXRs. The concatenated features are then fed to the fully connected (FC) layers for performing multi‐class categorization using Softmax activation function. The performance of proposed VEntNet model is compared with other DCNNs with and without the hybrid approach for categorization of closely related lung pathologies and normal CXR images. Our proposed VEntNet achieved accuracies of 98.78% and 90.96%, respectively, for four and five‐class classification of CXRs. Thus, it is demonstrated that among the different DCNNs, our VEntNet outperformed in four‐class CXR categorization tasks. The proposed model can potentially save time by facilitating the screening of CXRs to identify those with abnormalities present as well as to categorize the abnormalities. [ABSTRACT FROM AUTHOR]

Summary: Various deep neural network (DNN) model watermarks have been proposed by researchers to verify copyrights for deep neural networks DNN. However, most DNN watermarking methods cannot prevent attackers from stealing and using the model. Unlike many existing approaches, this paper uses a channel pruning algorithm to protect DNN models, which verifies DNN models copyrights but also prevents the illegal use of DNN models. In this work, the pruning threshold or pruning rate is used as the secret key of a DNN model. After the secret key is distributed to multiple users, they prune the DNN model with the secret key, and the pruned and fine‐tuned model is provided to the users. The users can verify ownership of the model according to the pruning accuracy and fine‐tuning accuracy. If the secret key is incorrect, the accuracy of the model after fine‐tuning will be very low, and users will be unable to use the reasoning function of the fine‐tuned model. Based on the CIFAR‐10 and CIFAR‐100 datasets, we conducted experiments on five popular DNN models. The experimental results show that we can authorize multiple users by pruning very few channels in the convolution layers of the DNN model. [ABSTRACT FROM AUTHOR]

Traditional Eulerian-based fluid simulations require much time and computational resources to solve the projection step, especially the large linear system produced by the Poisson equation. In this paper, we propose an adaptive machine-learning-based projection method combining deep neural network and incremental learning technique. We provide two modes: Fast Mode and Normal Mode to solve the most time-consuming projection step and deal with various simulation scenes largely different from the training data set. We introduce patch-based feature vectors to represent the whole velocity field and a loss function to keep the divergence-free constraint. We have demonstrated the acceleration and extrapolation ability of our method by testing various smoke scenes far different from our training data set. [ABSTRACT FROM AUTHOR]

Neural networks have generated valuable Quantitative Structure-Activity/Property Relationships (QSAR/QSPR) models for a wide variety of small molecules and materials properties. They have grown in sophistication and many of their initial problems have been overcome by modern mathematical techniques. QSAR studies have almost always used so-called 'shallow' neural networks in which there is a single hidden layer between the input and output layers. Recently, a new and potentially paradigm-shifting type of neural network based on Deep Learning has appeared. Deep learning methods have generated impressive improvements in image and voice recognition, and are now being applied to QSAR and QSAR modelling. This paper describes the differences in approach between deep and shallow neural networks, compares their abilities to predict the properties of test sets for 15 large drug data sets (the kaggle set), discusses the results in terms of the Universal Approximation theorem for neural networks, and describes how DNN may ameliorate or remove troublesome 'activity cliffs' in QSAR data sets. [ABSTRACT FROM AUTHOR]

Satellite‐based remote sensing missions have revolutionized our understanding of the Ocean state and dynamics. Among them, space‐borne altimetry provides valuable Sea Surface Height (SSH) measurements, used to estimate surface geostrophic currents. Due to the sensor technology employed, important gaps occur in SSH observations. Complete SSH maps are produced using linear Optimal Interpolations (OI) such as the widely used Data Unification and Altimeter Combination System (duacs). On the other hand, Sea Surface Temperature (SST) products have much higher data coverage and SST is physically linked to geostrophic currents through advection. We propose a new multi‐variate Observing System Simulation Experiment (OSSE) emulating 20 years of SSH and SST satellite observations. We train an Attention‐Based Encoder‐Decoder deep learning network (abed) on this data, comparing two settings: one with access to ground truth during training and one without. On our OSSE, we compare abed reconstructions when trained using either supervised or unsupervised loss functions, with or without SST information. We evaluate the SSH interpolations in terms of eddy detection. We also introduce a new way to transfer the learning from simulation to observations: supervised pre‐training on our OSSE followed by unsupervised fine‐tuning on satellite data. Based on real SSH observations from the Ocean Data Challenge 2021, we find that this learning strategy, combined with the use of SST, decreases the root mean squared error by 24% compared to OI. Plain Language Summary: The surface of the ocean is observed through various sensors embedded in satellites. Specifically, the height of the sea surface is a crucial variable as it can be used to estimate surface currents. It is currently measured through satellite altimeters, but the data acquisition process leaves gaps in their observations. Providing fully gridded maps of the sea surface height is thus an important interpolation problem. The widely used interpolated product has some troubles, especially when dealing with small and rapidly evolving eddies. To enhance its quality, we propose an artificial neural network, a trainable method able to estimate complete sea surface height images. The flexibility of these methods allows us to use different satellite information, such as the sea surface temperature, which has a better resolution. Usually, neural networks are trained on a data set upon which they learn the link between input and output data. However, in a realistic geoscience scenario, the output is never known for sure. We propose a data set that simulates this problem and explores methodologies to train these methods. Finally, we design a new way of learning SSH reconstruction using both simulated data and real satellite observation. Key Points: We developed a realistic simulation of satellite observations of sea surface height and temperatureWe compare deep learning supervised and unsupervised strategies to interpolate the sea surface heightWe present a hybrid learning strategy using supervised training on our simulated data and unsupervised fine‐tuning on real‐world data [ABSTRACT FROM AUTHOR]

In this paper, we propose a rank‐weighted reconstruction feature to improve the robustness of a feed‐forward deep neural network (FFDNN)‐based acoustic model. In the FFDNN‐based acoustic model, an input feature is constructed by vectorizing a submatrix that is created by slicing the feature vectors of frames within a context window. In this type of feature construction, the appropriate context window size is important because it determines the amount of trivial or discriminative information, such as redundancy, or temporal context of the input features. However, we ascertained whether a single parameter is sufficiently able to control the quantity of information. Therefore, we investigated the input feature construction from the perspectives of rank and nullity, and proposed a rank‐weighted reconstruction feature herein, that allows for the retention of speech information components and the reduction in trivial components. The proposed method was evaluated in the TIMIT phone recognition and Wall Street Journal (WSJ) domains. The proposed method reduced the phone error rate of the TIMIT domain from 18.4% to 18.0%, and the word error rate of the WSJ domain from 4.70% to 4.43%. [ABSTRACT FROM AUTHOR]

Summary: The objective of this paper is twofold. First, it develops a prediction system to help the credit card issuer model the credit card delinquency risk. Second, it seeks to explore the potential of deep learning (also called a deep neural network), an emerging artificial intelligence technology, in the credit risk domain. With real‐life credit card data linked to 711,397 credit card holders from a large bank in Brazil, this study develops a deep neural network to evaluate the risk of credit card delinquency based on the client's personal characteristics and the spending behaviours. Compared with machine‐learning algorithms of logistic regression, naive Bayes, traditional artificial neural networks, and decision trees, deep neural networks have a better overall predictive performance with the highest F scores and area under the receiver operating characteristic curve. The successful application of deep learning implies that artificial intelligence has great potential to support and automate credit risk assessment for financial institutions and credit bureaus. [ABSTRACT FROM AUTHOR]

It has been reported that training deep neural networks is more difficult than training shallow neural networks. Hinton et al. proposed deep belief networks with a learning algorithm that trains one layer at a time. A much better generalization can be achieved when pre‐training each layer with an unsupervised learning algorithm. Since then, deep neural networks have been extensively studied. On the other hand, it has been revealed that singular points affect the training dynamics of the learning models such as neural networks and cause a standstill of training. Naturally, training deep neural networks suffer singular points. As described in this paper, we present a deep neural network model that has fewer singular points than the usual one. First, we demonstrate that some singular points in the deep real‐valued neural network, which is equivalent to a deep complex‐valued neural network, have been resolved as its inherent property. Such deep neural networks are less likely to become trapped in local minima or plateaus caused by critical points. Results of experiments on the two spirals problem, which has an extreme nonlinearity, support our theory. Copyright © 2017 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Online training of deep neural networks (DNN) can be significantly accelerated by performing in situ vector‐matrix multiplication in a crossbar array of analog memories. However, training accuracies often suffer due to nonideal properties of synapses such as nonlinearity, asymmetry, limited bit precision, and dynamic weight update range within a constrained power budget. Herein, a fully scalable process is reported for digital and analog ferroelectric memory transistors with possibilities for both volatile and nonvolatile data retention and <4 V operation that would be suitable as programmable synaptic weight elements. Ferroelectric copolymer P(VDF‐TrFE) gate insulator and 2D semiconductor MoS2 as the n‐type semiconducting channel material make them suitable for flexible and wearable substrate integration. The ferroelectric‐only devices show excellent performance as digital nonvolatile memory operating at <±5 V while the hybrid ferroelectric–dielectric devices show quasi‐continuous resistive switching resulting from gradual ferroelectric domain rotation. Analog conductance states of the hybrid devices allow good linearity and symmetry of weight updates and produce a dynamic conductance range of 104 with >16 reproducible conducting states. Network training experiments with these ferroelectric field‐effect transistors show >96% classification accuracy with Modified National Institute of Standards and Technology (MNIST) handwritten datasets highlighting their potential for implementation in scaled DNN architectures. [ABSTRACT FROM AUTHOR]

Despite great progress in seeking accurate numerical approximator to nonlinear structural seismic response prediction using deep learning approaches, tedious training process and large volume of structural response data under earthquakes for training and validation are often prohibitively accessible. In our methodology, the main innovation can be seen in the incorporation of deep neural networks (DNNs) into a classical numerical integration method by using a hybridized integration time‐stepper. In this way, the linear physics information of the structure and the obscure nonlinear dynamics are smoothly combined. We propose to use residual network (ResNet) to learn time‐stepping schemes specifically for the nonlinear state variables of the system. Our Physics‐DNN hybridized integration (PDHI) time‐stepping scheme provides important advantages over current pure data‐driven approaches, including (i) a flexible framework incorporating known time‐invariant physics information, (ii) requirement of structural seismic response data being circumvented by simple short bursts of trajectories collected from underlying nonlinear components, and (iii) efficiency in training and validation process. Besides, our results indicate that a simple feedforward or convolutional architecture outperforms recurrent networks to fulfill the requirement of prediction accuracy as well as long‐range memory in structural dynamic analysis. Several numerical and experimental examples are presented to demonstrate the performance of the method. [ABSTRACT FROM AUTHOR]

In engineering systems such as energy systems and electronic devices, thermal reliability analysis of the heat source component area is critical to ensure the proper functioning of these components. Previous methods, due to the high data dimensionality of the temperature field of the whole heat source area, face high computational cost when analyzing the reliability. To solve the problem of high computational cost, reliability analysis methods based on deep neural network surrogate models are receiving more and more attention. However, the uncertainty of deep neural networks is rarely considered in reliability problems, leading to unreasonable or imperfect reliability analysis. Therefore, we propose a reliability analysis method for heat source layout temperature field prediction, considering the uncertainty of deep neural network surrogate models. The method uses the information on the standard deviation of temperature field prediction provided by the model uncertainty to obtain the interval prediction of the temperature field. Then, the temperature field reliability interval of the layout area can be obtained by combining the temperature field interval prediction and Monte Carlo sampling. Further, we propose a strategy for selecting the heat source layout scheme based on the reliability interval in the late engineering design stage. Finally, the method's validity is verified using some examples, which provides an important reference basis for engineers. [ABSTRACT FROM AUTHOR]

Next‐generation displays require tiny pixel sizes and are applied to AR/VR and holograms. Conventional color filters have limitations in miniaturization and shows poor reliability from heat, light and other environment. Meta‐surface color filters (MCFs) are one of the solutions to replace conventional color filters, suitable for high‐resolution displays. Structural design parameters are adjusted to determine the filtering properties of the MCFs. In addition, the metal based meta‐surface shows better durability cause of the intrinsic characteristics of the material. However, if the structure is complicated, there require many case‐ studies for verification, which increases trial and error. In addition, resonance by phase matching between structures changes according to the angle of incident light. In this study, we suggest a method to reduce the trial and error required in designing structure. Also, the design parameters of the MCFs of the desired color was extracted through an AI model capable of reverse design. Furthermore, we propose a novel MCFs that shows characteristics independent to the angle of incident light through a coaxial hole and disk array structure. [ABSTRACT FROM AUTHOR]

Phase change memory (PCM) is one of the most promising candidates for non‐von Neumann based analog in‐memory computing–particularly for inference of previously‐trained deep neural networks (DNN). It is shown that PCM electrical properties can be tuned systematically using a projection liner, which is designed for resistance drift mitigation, in the manufacturable mushroom PCM. A systematic study of the electrical properties‐including resistance values, memory window, resistance drift, read noise, and their impact on the accuracy of large neural networks of various types and with tens of millions of weights is performed. It is sown that the DNN accuracy can be improved by the PCM with liner for both the short term and long term after programming, due to reduced resistance drift and read noise, respectively, despite the trade‐off of reduced memory window. The liner conductance, PCM device characteristics, and network inference accuracy with PCM memory window and reset state conductance is correlated, which allows us to identify the device optimization space to achieve better short term and long term accuracy for large neural networks. [ABSTRACT FROM AUTHOR]

Phishing attacks have become one of the powerful sources for cyber criminals to impose various forms of security attacks in which fake website Uniform Resource Locators (URL) are circulated around the Internet community in the form of email, messages etc., in order to deceive users, resulting in the loss of their valuable assets. The phishing URLs are predicted using several blacklist‐based traditional phishing website detection techniques. However, numerous phishing websites are frequently constructed and launched on the Internet over time; these blacklist‐based traditional methods do not accurately predict most phishing websites. In order to effectively identify malicious URLs, an enhanced deep learning‐based phishing detection approach has been proposed by integrating the strength of Variational Autoencoders (VAE) and deep neural networks (DNN). In the proposed framework, the inherent features of a raw URL are automatically extracted by the VAE model by reconstructing the original input URL to enhance phishing URL detection. For experimentation, around 1 lakh URLs were crawled from two publicly available datasets, namely ISCX‐URL‐2016 dataset and Kaggle dataset. The experimental results suggested that the proposed model has reached a maximum accuracy of 97.45% and exhibits a quicker response time of 1.9 s, which is better when compared to all the other experimented models. [ABSTRACT FROM AUTHOR]

Previous studies have explored resting‐state functional connectivity (rs‐FC) of the amygdala in patients with autism spectrum disorder (ASD). However, it remains unclear whether there are frequency‐specific FC alterations of the amygdala in ASD and whether FC in specific frequency bands can be used to distinguish patients with ASD from typical controls (TCs). Data from 306 patients with ASD and 314 age‐matched and sex‐matched TCs were collected from 28 sites in the Autism Brain Imaging Data Exchange database. The bilateral amygdala, defined as the seed regions, was used to perform seed‐based FC analyses in the conventional, slow‐5, and slow‐4 frequency bands at each site. Image‐based meta‐analyses were used to obtain consistent brain regions across 28 sites in the three frequency bands. By combining generative adversarial networks and deep neural networks, a deep learning approach was applied to distinguish patients with ASD from TCs. The meta‐analysis results showed frequency band specificity of FC in ASD, which was reflected in the slow‐5 frequency band instead of the conventional and slow‐4 frequency bands. The deep learning results showed that, compared with the conventional and slow‐4 frequency bands, the slow‐5 frequency band exhibited a higher accuracy of 74.73%, precision of 74.58%, recall of 75.05%, and area under the curve of 0.811 to distinguish patients with ASD from TCs. These findings may help us to understand the pathological mechanisms of ASD and provide preliminary guidance for the clinical diagnosis of ASD. [ABSTRACT FROM AUTHOR]

Summary: The easiest way to propose a solution based on deep neural networks is using the pre‐trained models through the transfer‐learning technique. Deep learning platforms provide various pre‐trained deep neural networks that can be easily applied for image classification tasks. So, "Which pre‐trained model provides the best performance for image classification tasks?" is a question that instinctively comes to mind and should be shed light on by the research community. To this end, we propose an experimental comparison of the six popular pre‐trained deep neural networks, namely, (i) VGG19, (ii) ResNet50, (iii) DenseNet201, (iv) MobileNetV2, (v) InceptionV3, and (vi) Xception by employing them through the transfer‐learning technique. Then, the proposed benchmark models were both trained and evaluated under the same configurations on two gold‐standard datasets, namely, (i) CIFAR‐10 and (ii) Stanford Dogs to benchmark them. Three evaluation metrics were employed to measure performance differences between the employed pre‐trained models as follows: (i) Accuracy, (ii) training duration, and (iii) inference time. The key findings that were obtained through the conducted a wide variety of experiments were discussed. [ABSTRACT FROM AUTHOR]

Extraction of intravoxel incoherent motion (IVIM) parameters from noisy diffusion‐weighted (DW) images using a biexponential fitting model is computationally challenging, and the reliability of the estimated perfusion‐related quantities represents a limitation of this technique. Artificial intelligence can overcome the current limitations and be a suitable solution to advance use of this technique in both preclinical and clinical settings. The purpose of this work was to develop a deep neural network (DNN) approach, trained on numerical simulated phantoms with different signal to noise ratios (SNRs), to improve IVIM parameter estimation. The proposed approach is based on a supervised fully connected DNN having 3 hidden layers, 18 inputs and 3 targets with standardized values. 14 × 103 simulated DW images, based on a Shepp–Logan phantom, were randomly generated with varying SNRs (ranging from 10 to 100). 7 × 103 images (1000 for each SNR) were used for training. Performance accuracy was assessed in simulated images and the proposed approach was compared with the state‐of‐the‐art Bayesian approach and other DNN algorithms. The DNN approach was also evaluated in vivo on a high‐field MRI preclinical scanner. Our DNN approach showed an overall improvement in accuracy when compared with the Bayesian approach and other DNN methods in most of the simulated conditions. The in vivo results demonstrated the feasibility of the proposed approach in real settings and generated quantitative results comparable to those obtained using the Bayesian and unsupervised approaches, especially for D and f, and with lower variability in homogeneous regions. The DNN architecture proposed in this work outlines two innovative features as compared with other studies: (1) the use of standardized targets to improve the estimation of parameters, and (2) the implementation of a single DNN to enhance the IVIM fitting at different SNRs, providing a valuable alternative tool to compute IVIM parameters in conditions of high background noise. [ABSTRACT FROM AUTHOR]

Geosteering of wells requires fast interpretation of geophysical logs which is a non‐unique inverse problem. Current work presents a proof‐of‐concept approach to multi‐modal probabilistic inversion of logs using a single evaluation of an artificial deep neural network (DNN). A mixture density DNN (MDN) is trained using the "multiple‐trajectory‐prediction" loss functions, which avoids mode collapse typical for traditional MDNs, and allows multi‐modal prediction ahead of data. The proposed approach is verified on the real‐time stratigraphic inversion of gamma‐ray logs. The multi‐modal predictor outputs several likely inverse solutions/predictions, providing more accurate and realistic solutions compared to a deterministic regression using a DNN. For these likely stratigraphic curves, the model simultaneously predicts their probabilities, which are implicitly learned from the training geological data. The stratigraphy predictions and their probabilities obtained in milliseconds from the MDN can enable better real‐time decisions under geological uncertainties. Plain Language Summary: Positioning the wells relative to geological targets and adjusting trajectory in real‐time requires fast interpretation of streamed geophysical measurements. As such interpretations are not unique, high‐quality decision‐making requires the exploration of all likely interpretations and estimation of their probabilities. This study presents a mixture density deep neural network that correlates the log of the drilled well with the offset well and outputs a chosen number of interpretations of the geometry of geological layers and their probabilities. Moreover, by learning the likely configurations in the training geological data, one can extrapolate the interpretations ahead of the data. The presented model achieves good prediction accuracy while producing more realistic interpretations compared to the deterministic single‐output model. Key Points: Multi‐modal probabilistic inversion of geophysical logs using a single evaluation of a deep neural networkDeep neural network outputs likely stratigraphic inversions and predictions ahead of data and their probabilitiesThe model predicts more accurate and realistic solutions compared to the single‐mode predictor [ABSTRACT FROM AUTHOR]

Summary: Breast cancer detection is the most challenging aspect in the field of health monitoring system. In this paper, breast cancer detection was assessed by employing Mammographic Image Analysis Society (MIAS) dataset. The proposed approach contains four major steps, namely, image‐preprocessing, segmentation, feature extraction, and classification. Initially, Laplacian filtering was utilized to identify the area of edges in mammogram images and, also, it was very sensitive to noise. Then, segmentation was carried‐out using modified‐Adaptively Regularized Kernel‐based Fuzzy‐C‐Means (ARKFCM); it was a flexible high level machine learning technique to localize the object in complex template. In conventional ARKFCM, it was hard to segment the ill‐defined masses in mammogram images. To address this concern, the Euclidean distance in ARKFCM was replaced by correlation function in order to improve the segmentation efficiency. The hybrid feature extraction (Histogram of Oriented Gradients (HOG), homogeneity, and energy) was performed on the segmented cancer region to extract feature subsets. The respective feature values were given as the input for a multi‐objective classifier: Deep Neural Network (DNN) for classifying the normal and abnormal regions in mammogram images. The experimental outcome shows that the proposed methodology improved accuracy in breast cancer classification up to 3% to 9% compared to other existing methods. [ABSTRACT FROM AUTHOR]

Wildfires not only severely damage the natural environment and global ecological balance but also cause substantial losses to global forest resources and human lives and property. Unprecedented fire events such as Australia's bushfires have alerted us to the fact that wildfire prediction is a critical scientific problem for fire management. Therefore, robust, long‐lead models and dynamic predictions of wildfire are valuable for global fire prevention. However, despite decades of effort, the dynamic, effective, and accurate prediction of wildfire remains problematic. There is great uncertainty in predicting the future based on historical and existing spatiotemporal sequence data, but with advances in deep learning algorithms, solutions to prediction problems are being developed. Here, we present a dynamic prediction model of global burned area of wildfire employing a deep neural network (DNN) approach that produces effective wildfire forecasts based on historical time series predictors and satellite‐based burned area products. A hybrid DNN that combines long short‐term memory and a two‐dimensional convolutional neural network (CNN2D‐LSTM) was proposed, and CNN2D‐LSTM model candidates with four different architectures were designed and compared to construct the optimal architecture for fire prediction. The proposed model was also shown to outperform convolutional neural networks (CNNs) and the fully connected long short‐term memory (FcLSTM) approach using the refined index of agreement and evaluation metrics. We produced monthly global burned area spatiotemporal prediction maps and adequately reflected the seasonal peak in fire activity and highly fire‐prone areas. Our combined CNN2D‐LSTM approach can effectively predict the global burned area of wildfires 1 month in advance and can be generalized to provide seasonal estimates of global fire risk. [ABSTRACT FROM AUTHOR]

Compute is a term coined from the etymology of French and Latin words computer and computare respectively, so is computing. This field of computing has grown enormously over the years. From the simple, traditional Turing machine invented in 1936 by Alan Turing to the current neural network (NN) computing. NNs, a field of artificial intelligence (AI) was exhilarated from the structure and inner workings of the brain. Just as the brain is, that is, an interconnection of neurons, so is the NN which is an interconnection of basic structures known as the perceptron. They do not differ much in structure. Their only difference is that one is artificial while the other is entirely biological. The hierarchical intricacies of the NN can be represented in three layers: the perceptron, artificial NN (ANN), and deep NN (DNN). With the influx of mental and behavioral disorders, basic surveillance, and the urgency to improve the mental health of people, studying the behavioral dynamics of people is requisite. CCTV and street cameras can only do so much, thus the need to employ the field of NN which makes use of supervised learning in training the models to perfect and automate surveillance. The results of this retrospective research indicate that the use of the NN model surpasses those of traditional methods in terms of efficiency and reliability. [ABSTRACT FROM AUTHOR]

Diabetic retinopathy (DR) accounts in eye‐related disorders due to accumulated damage to small retinal blood vessels. Automated diagnostic systems are effective in early detection and diagnosis of severe eye complications by assisting the ophthalmologists. Deep learning‐based techniques have emerged as an advancement over conventional techniques based on hand‐crafted features. The authors have proposed a Quadrant‐based automated DR grading system in this work using Inception‐V3 deep neural network to extract small lesions present in retinal fundus images. The grading efficiency of the proposed architecture is improved utilizing image enhancement and optical disc removal pipeline along with data augmentation stage. The proposed system yields accuracy of 93.33% with minimized cross‐entropy loss of 0.291. Capability of proposed system is demonstrated experimentally to provide efficient DR diagnosis. The diagnosis ability of the proposed architecture is demonstrated by state‐of‐the‐art comparison with other mainstream convolution neural network models and a maximum improvement of 14.33% is observed. [ABSTRACT FROM AUTHOR]

Since the QSAR/DNN model showed predominant predictive performance over other conventional methods in the Kaggle QSAR competition, many artificial neural network (ANN) methods have been applied to drug and material discovery. Appearance of artificial intelligence (AI), which is combined various general purpose ANN platforms with large‐scale open access chemical databases, has attracting great interest and expectation in a wide range of molecular sciences. In this study, we investigate various DNN settings in order to reach a high‐level of predictive performance comparable to the champion team of the competition, even with a general purpose ANN platform, and introduce the Meister setting for constructing a good QSAR/DNNs model. Here, we have used the most commonly available DNN model and constructed many QSAR/DNN models trained with various DNN settings by using the 15 datasets employed in the competition. As a result, it was confirmed that we can constructed the QSAR/DNN model that shows the same level of R2 performance as the champion team. The difference from the DNN setting recommended by the champion team was to reduce the mini‐batch size. We have also explained that the R2 performance of each target depends on the molecular activity type, which is related to the complexity of biological mechanisms and chemical processes observed in molecular activity measurements. [ABSTRACT FROM AUTHOR]