Search

RNA-binding proteins (RBPs) play an important role in regulating biological processes, such as gene regulation. Understanding their behaviors, for example, their binding site, can be helpful in understanding RBP-related diseases. Studies have focused on predicting RNA binding by means of machine learning algorithms including deep convolutional neural network models. One of the integral parts of modeling deep learning is achieving optimal hyperparameter tuning and minimizing a loss function using optimization algorithms. In this paper, we investigate the role of optimization in the RBP classification problem using the CLIP-Seq 21 dataset. Three optimization methods are employed on the RNA–protein binding CNN prediction model; namely, grid search, random search, and Bayesian optimizer. The empirical results show an AUC of 94.42%, 93.78%, 93.23% and 92.68% on the ELAVL1C, ELAVL1B, ELAVL1A, and HNRNPC datasets, respectively, and a mean AUC of 85.30 on 24 datasets. This paper’s findings provide evidence on the role of optimizers in improving the performance of RNA–protein binding prediction.

Deep learning, a crucial technique for achieving artificial intelligence (AI), has been successfully applied in many fields. The gradual application of the latest architectures of deep learning in the field of time series forecasting (TSF), such as Transformers, has shown excellent performance and results compared to traditional statistical methods. These applications are widely present in academia and in our daily lives, covering many areas including forecasting electricity consumption in power systems, meteorological rainfall, traffic flow, quantitative trading, risk control in finance, sales operations and price predictions for commercial companies, and pandemic prediction in the medical field. Deep learning-based TSF tasks stand out as one of the most valuable AI scenarios for research, playing an important role in explaining complex real-world phenomena. However, deep learning models still face challenges: they need to deal with the challenge of large-scale data in the information age, achieve longer forecasting ranges, reduce excessively high computational complexity, etc. Therefore, novel methods and more effective solutions are essential. In this paper, we review the latest developments in deep learning for TSF. We begin by introducing the recent development trends in the field of TSF and then propose a new taxonomy from the perspective of deep neural network models, comprehensively covering articles published over the past five years. We also organize commonly used experimental evaluation metrics and datasets. Finally, we point out current issues with the existing solutions and suggest promising future directions in the field of deep learning combined with TSF. This paper is the most comprehensive review related to TSF in recent years and will provide a detailed index for researchers in this field and those who are just starting out.

Utilizing multi-modal data, as opposed to only hyperspectral image (HSI), enhances target identification accuracy in remote sensing. Transformers are applied to multi-modal data classification for their long-range dependency but often overlook intrinsic image structure by directly flattening image blocks into vectors. Moreover, as the encoder deepens, unprofitable information negatively impacts classification performance. Therefore, this paper proposes a learnable transformer with an adaptive gating mechanism (AGMLT). Firstly, a spectral–spatial adaptive gating mechanism (SSAGM) is designed to comprehensively extract the local information from images. It mainly contains point depthwise attention (PDWA) and asymmetric depthwise attention (ADWA). The former is for extracting spectral information of HSI, and the latter is for extracting spatial information of HSI and elevation information of LiDAR-derived rasterized digital surface models (LiDAR-DSM). By omitting linear layers, local continuity is maintained. Then, the layer Scale and learnable transition matrix are introduced to the original transformer encoder and self-attention to form the learnable transformer (L-Former). It improves data dynamics and prevents performance degradation as the encoder deepens. Subsequently, learnable cross-attention (LC-Attention) with the learnable transfer matrix is designed to augment the fusion of multi-modal data by enriching feature information. Finally, poly loss, known for its adaptability with multi-modal data, is employed in training the model. Experiments in the paper are conducted on four famous multi-modal datasets: Trento (TR), MUUFL (MU), Augsburg (AU), and Houston2013 (HU). The results show that AGMLT achieves optimal performance over some existing models.

The cause mechanism of collapse disasters is complex and there are many influencing factors. Convolutional Neural Network (CNN) has a strong feature extraction ability, which can better simulate the formation of collapse disasters and accurately predict them. Taihe town’s collapse threatens roads, buildings, and people. In this paper, road distance, water distance, normalized vegetation index, platform curvature, profile curvature, slope, slope direction, and geological data are used as input variables. This paper generates collapse susceptibility zoning maps based on the information value method (IV) and CNN, respectively. The results show that the accuracy of the susceptibility assessment of the IV method and the CNN method is 85.1% and 87.4%, and the accuracy of the susceptibility assessment based on the CNN method is higher. The research results can provide some reference for the formulation of disaster prevention and control strategies.

Hyperspectral images’ (HSIs) classification research has seen significant progress with the use of convolutional neural networks (CNNs) and Transformer blocks. However, these studies primarily incorporated Transformer blocks at the end of their network architectures. Due to significant differences between the spectral and spatial features in HSIs, the extraction of both global and local spectral–spatial features remains incomplete. To address this challenge, this paper introduces a novel method called TransHSI. This method incorporates a new spectral–spatial feature extraction module that leverages 3D CNNs to fuse Transformer to extract the local and global spectral features of HSIs, then combining 2D CNNs and Transformer to capture the local and global spatial features of HSIs comprehensively. Furthermore, a fusion module is proposed, which not only integrates the learned shallow and deep features of HSIs but also applies a semantic tokenizer to transform the fused features, enhancing the discriminative power of the features. This paper conducts experiments on three public datasets: Indian Pines, Pavia University, and Data Fusion Contest 2018. The training and test sets are selected based on a disjoint sampling strategy. We perform a comparative analysis with 11 traditional and advanced HSI classification algorithms. The experimental results demonstrate that the proposed method, TransHSI algorithm, achieves the highest overall accuracies and kappa coefficients, indicating a competitive performance.

In order to more effectively monitor and interfere with enemy signals, it is particularly important to accurately and efficiently identify the intercepted signals and estimate their parameters in the increasingly complex electromagnetic environment. Therefore, in non-cooperative situations, it is of great practical significance to study how to accurately detect direct sequence spread spectrum (DSSS) signals in real time and estimate their parameters. The traditional time-delay correlation algorithm encounters the challenges such as peak energy leakage and false peak interference. As an alternative, this paper introduces a Pseudo-Noise (PN) code period estimation method utilizing a one-dimensional (1D) convolutional neural network based on the residual network (CNN-ResNet). This method transforms the problem of spread spectrum code period estimation into a multi-classification problem of spread spectrum code length estimation. Firstly, the In-phase/Quadrature(I/Q) two-way of the received DSSS signals is directly input into the CNN-ResNet model, which will automatically learn the characteristics of the DSSS signal with different PN code lengths and then estimate the PN code length. Simulation experiments are conducted using a data set with DSSS signals ranging from −20 to 10 dB in terms of signal-to-noise ratios (SNRs). Upon training and verifying the model using BPSK modulation, it is then put to the test with QPSK-modulated signals, and the estimation performance was analyzed through metrics such as loss function, accuracy rate, recall rate, and confusion matrix. The results demonstrate that the 1D CNN-ResNet proposed in this paper is capable of effectively estimating the PN code period of the non-cooperative DSSS signal, exhibiting robust generalization abilities.

With the wide application of direct sequence spread spectrum (DSSS) signals, the comprehensive performance of DSSS communication systems has been continuously improved, making the electronic reconnaissance link in communication countermeasures more difficult. Electronic reconnaissance technology, as the fundamental means of modern electronic warfare, mainly includes signal detection, recognition, and parameter estimation. At present, research on DSSS detection algorithms is mostly based on the correlation characteristics of DSSS signals, and autocorrelation algorithm is the most mature and widely used method in practical engineering. With the continuous development of deep learning, deep-learning-based methods have gradually been introduced to replace traditional algorithms in the field of signal processing. This paper proposes a spread spectrum signal detection method based on convolutional neural network (CNN). Through experimental analysis, the detection performance of the CNN model proposed in this paper on DSSS signals in various situations has been compared and analyzed with traditional autocorrelation detection methods for different signal-to-noise ratios. The experiments verified the estimation performance of the model in this paper under different signal-to-noise ratios, different spreading code lengths, different spreading code types, and different modulation methods and compared it with the autocorrelation detection algorithm. It was found that the detection performance of the model in this paper was higher than that of the autocorrelation detection method, and the overall performance was improved by 4 dB.

In the field of image fusion, spatial detail blurring and color distortion appear in synthetic aperture radar (SAR) images and multispectral (MS) during the traditional fusion process due to the difference in sensor imaging mechanisms. To solve this problem, this paper proposes a fusion method for SAR images and MS images based on a convolutional neural network. In order to make use of the spatial information and different scale feature information of high-resolution SAR image, a dual-channel feature extraction module is constructed to obtain a SAR image feature map. In addition, different from the common direct addition strategy, an attention-based feature fusion module is designed to achieve spectral fidelity of the fused images. In order to obtain better spectral and spatial retention ability of the network, an unsupervised joint loss function is designed to train the network. In this paper, the Sentinel 1 SAR images and Landsat 8 MS images are used as datasets for experiments. The experimental results show that the proposed algorithm has better performance in quantitative and visual representation when compared with traditional fusion methods and deep learning algorithms.

Breast cancer poses the greatest long-term health risk to women worldwide, in both industrialized and developing nations. Early detection of breast cancer allows for treatment to begin before the disease has a chance to spread to other parts of the body. The Internet of Things (IoT) allows for automated analysis and classification of medical pictures, allowing for quicker and more effective data processing. Nevertheless, Fog computing principles should be used instead of Cloud computing concepts alone to provide rapid responses while still meeting the requirements for low latency, energy consumption, security, and privacy. In this paper, we present CanDiag, an approach to cancer diagnosis based on Transfer Deep Learning (TDL) that makes use of Fog computing. This paper details an automated, real-time approach to diagnosing breast cancer using deep learning (DL) and mammography pictures from the Mammographic Image Analysis Society (MIAS) library. To obtain better prediction results, transfer learning (TL) techniques such as GoogleNet, ResNet50, ResNet101, InceptionV3, AlexNet, VGG16, and VGG19 were combined with the well-known DL approach of the convolutional neural network (CNN). The feature reduction technique principal component analysis (PCA) and the classifier support vector machine (SVM) were also applied with these TDLs. Detailed simulations were run to assess seven performance and seven network metrics to prove the viability of the proposed approach. This study on an enormous dataset of mammography images categorized as normal and abnormal, respectively, achieved an accuracy, MCR, precision, sensitivity, specificity, f1-score, and MCC of 99.01%, 0.99%, 98.89%, 99.86%, 95.85%, 99.37%, and 97.02%, outperforming some previous studies based on mammography images. It can be shown from the trials that the inclusion of the Fog computing concepts empowers the system by reducing the load on centralized servers, increasing productivity, and maintaining the security and integrity of patient data.

The application of Transformer in computer vision has had the most significant influence of all the deep learning developments over the past five years. In addition to the exceptional performance of convolutional neural networks (CNN) in hyperspectral image (HSI) classification, Transformer has begun to be applied to HSI classification. However, for the time being, Transformer has not produced satisfactory results in HSI classification. Recently, in the field of image classification, the creators of Sequencer have proposed a Sequencer structure that substitutes the Transformer self-attention layer with a BiLSTM2D layer and achieves satisfactory results. As a result, this paper proposes a unique network called SquconvNet, that combines CNN with Sequencer block to improve hyperspectral classification. In this paper, we conducted rigorous HSI classification experiments on three relevant baseline datasets to evaluate the performance of the proposed method. The experimental results show that our proposed method has clear advantages in terms of classification accuracy and stability.

Due to the imaging mechanism of synthetic aperture radar (SAR), it is difficult and costly to acquire abundant labeled SAR images. Moreover, a typical matched filtering (MF) based image faces the problems of serious noise, sidelobes, and clutters, which will bring down the accuracy of SAR target classification. Different from the MF-based result, a sparse image shows better quality with less noise and higher image signal-to-noise ratio (SNR). Therefore, theoretically using it for target classification will achieve better performance. In this paper, a novel contrastive domain adaptation (CDA) based sparse SAR target classification method is proposed to solve the problem of insufficient samples. In the proposed method, we firstly construct a sparse SAR image dataset by using the complex image based iterative soft thresholding (BiIST) algorithm. Then, the simulated and real SAR datasets are simultaneously sent into an unsupervised domain adaptation framework to reduce the distribution difference and obtain the reconstructed simulated SAR images for subsequent target classification. Finally, the reconstructed simulated images are manually labeled and fed into a shallow convolutional neural network (CNN) for target classification along with a small number of real sparse SAR images. Since the current definition of the number of small samples is still vague and inconsistent, this paper defines few-shot as less than 20 per class. Experimental results based on MSTAR under standard operating conditions (SOC) and extended operating conditions (EOC) show that the reconstructed simulated SAR dataset makes up for the insufficient information from limited real data. Compared with other typical deep learning methods based on limited samples, our method is able to achieve higher accuracy especially under the conditions of few shots.

In this paper, aiming at the severe problems of UWB positioning in NLOS-interference circumstances, a complete method is proposed for NLOS/LOS classification, NLOS identification and mitigation, and a final accurate UWB coordinate solution through the integration of two machine learning algorithms and a hybrid localization algorithm, which is called the C-T-CNN-SVM algorithm. This algorithm consists of three basic processes: an LOS/NLOS signal classification method based on SVM, an NLOS signal recognition and error elimination method based on CNN, and an accurate coordinate solution based on the hybrid weighting of the Chan–Taylor method. Finally, the validity and accuracy of the C-T-CNN-SVM algorithm are proved through a comparison with traditional and state-of-the-art methods. (i) Focusing on four main prediction errors (range measurements, maxNoise, stdNoise and rangeError), the standard deviation decreases from 13.65 cm to 4.35 cm, while the mean error decreases from 3.65 cm to 0.27 cm, and the errors are practically distributed normally, demonstrating that after training a SVM for LOS/NLOS signal classification and a CNN for NLOS recognition and mitigation, the accuracy of UWB range measurements may be greatly increased. (ii) After target positioning, the proposed method can realize a one-dimensional X-axis and Y-axis accuracy within 175 mm, and a Z-axis accuracy within 200 mm; a 2D (X,Y) accuracy within 200 mm; and a 3D accuracy within 200 mm, most of which fall within (100 mm, 100 mm, 100 mm). (iii) Compared with the traditional algorithms, the proposed C-T-CNN-SVM algorithm performs better in location accuracy, cumulative error probability (CDF), and root-mean-square difference (RMSE): the 1D, 2D, and 3D accuracy of the proposed method is 2.5 times that of the traditional methods. When the location error is less than 10 cm, the CDF of the proposed algorithm only reaches a value of 0.17; when the positioning error reaches 30 cm, only the CDF of the proposed algorithm remains in an acceptable range. The RMSE of the proposed algorithm remains ideal when the distance error is greater than 30 cm. The results of this paper and the idea of a combination of machine learning methods with the classical locating algorithms for improved UWB positioning under NLOS interference could meet the growing need for wireless indoor locating and communication, which indicates the possibility for the practical deployment of such a method in the future.

Synthetic Aperture Radar (SAR) target detection is a significant research direction in radar information processing. Aiming at the poor robustness and low detection accuracy of traditional detection algorithms, SAR image target detection based on the Convolutional Neural Network (CNN) is reviewed in this paper. Firstly, the traditional SAR image target detection algorithms are briefly discussed, and their limitations are pointed out. Secondly, the CNN’s network principle, basic structure, and development process in computer vision are introduced. Next, the SAR target detection based on CNN is emphatically analyzed, including some common data sets and image processing methods for SAR target detection. The research status of SAR image target detection based on CNN is summarized and compared in detail with traditional algorithms. Afterward, the challenges of SAR image target detection are discussed and future research is proposed. Finally, the whole article is summarized. By summarizing and analyzing prior research work, this paper is helpful for subsequent researchers to quickly recognize the current development status and identify the connections between various detection algorithms. Beyond that, this paper summarizes the problems and challenges confronting researchers in the future, and also points out the specific content of future research, which has certain guiding significance for promoting the progress of SAR image target detection.

Cross-project defect prediction (CPDP) is a practical approach for finding software defects in projects which have incomplete or fewer data. Improvements to the defect prediction accuracy of CPDP—such as the PROMISE repository, the correct classification of the source data, removing the noise, reducing the distribution gap, and balancing the output classes—are an ongoing challenge, as is the selection of an optimal feature set. This research paper aims to achieve a higher defect prediction accuracy for multi-class CPDP by selecting an optimal feature set through XGBoost combined with an automatic feature extraction using a convolutional neural network (CNN). This research type is explanatory, and this research method is controlled experimentation, for which the independent variable prediction accuracy was dependent upon two variables, XGBoost and CNN. The Softmax layer was added to the output layers of the CNN classifier to classify the output into multiple classes. In our experimentation with CPDP, we selected all 28 versions of the multi-class, in which 11 versions were selected as the source projects, against which we predicted 28 target versions with an average AUC of 75.57%. We validated this research paper’s results through the Wilcoxon test. Therefore, after removing the noise, class imbalances, and the data distribution gap, and treating the PROMISE dataset as multi-class, the optimal features selected through XGBoost and classified through the CNN can substantially increase the prediction accuracy in CPDP as evident from our exploratory data analysis (EDA).

With the recent growth of the Internet of Things (IoT) and the demand for faster computation, quantized neural networks (QNNs) or QNN-enabled IoT can offer better performance than conventional convolution neural networks (CNNs). With the aim of reducing memory access costs and increasing the computation efficiency, QNN-enabled devices are expected to transform numerous industrial applications with lower processing latency and power consumption. Another form of QNN is the binarized neural network (BNN), which has 2 bits of quantized levels. In this paper, CNN-, QNN-, and BNN-based pattern recognition techniques are implemented and analyzed on an FPGA. The FPGA hardware acts as an IoT device due to connectivity with the cloud, and QNN and BNN are considered to offer better performance in terms of low power and low resource use on hardware platforms. The CNN and QNN implementation and their comparative analysis are analyzed based on their accuracy, weight bit error, RoC curve, and execution speed. The paper also discusses various approaches that can be deployed for optimizing various CNN and QNN models with additionally available tools. The work is performed on the Xilinx Zynq 7020 series Pynq Z2 board, which serves as our FPGA-based low-power IoT device. The MNIST and CIFAR-10 databases are considered for simulation and experimentation. The work shows that the accuracy is 95.5% and 79.22% for the MNIST and CIFAR-10 databases, respectively, for full precision (32-bit), and the execution time is 5.8 ms and 18 ms for the MNIST and CIFAR-10 databases, respectively, for full precision (32-bit).

Existing data-driven technology for prediction of state of health (SOH) has insufficient feature extraction capability and limited application scope. To deal with this challenge, this paper proposes a battery SOH prediction model based on multi-feature fusion. The model is based on a convolutional neural network (CNN) and a long short-term memory network (LSTM). The CNN can learn the cycle features in the battery data, the LSTM can learn the aging features of the battery over time, and regression prediction can be made through the full-connection layer (FC). In addition, for the aging differences caused by different battery operating conditions, this paper introduces transfer learning (TL) to improve the prediction effect. Across cycle data of the same battery under 12 different charging conditions, the fusion model in this paper shows higher prediction accuracy than with either LSTM and CNN in isolation, reducing RMSPE by 0.21% and 0.19%, respectively.

Diagnosing and screening for diabetic retinopathy is a well-known issue in the biomedical field. A component of computer-aided diagnosis that has advanced significantly over the past few years as a result of the development and effectiveness of deep learning is the use of medical imagery from a patient’s eye to identify the damage caused to blood vessels. Issues with unbalanced datasets, incorrect annotations, a lack of sample images, and improper performance evaluation measures have negatively impacted the performance of deep learning models. Using three benchmark datasets of diabetic retinopathy, we conducted a detailed comparison study comparing various state-of-the-art approaches to address the effect caused by class imbalance, with precision scores of 93%, 89%, 81%, 76%, and 96%, respectively, for normal, mild, moderate, severe, and DR phases. The analyses of the hybrid modeling, including CNN analysis and SHAP model derivation results, are compared at the end of the paper, and ideal hybrid modeling strategies for deep learning classification models for automated DR detection are identified.

Recognizing different contact patterns imposed on tactile sensors plays a very important role in human–machine interaction. In this paper, a flexible tactile sensor with great dynamic response characteristics is designed and manufactured based on polyvinylidene fluoride (PVDF) material. Four contact patterns (stroking, patting, kneading, and scratching) are applied to the tactile sensor, and time sequence data of the four contact patterns are collected. After that, a fusion model based on the convolutional neural network (CNN) and the long-short term memory (LSTM) neural network named CNN-LSTM is constructed. It is used to classify and recognize the four contact patterns loaded on the tactile sensor, and the recognition accuracies of the four patterns are 99.60%, 99.67%, 99.07%, and 99.40%, respectively. At last, a CNN model and a random forest (RF) algorithm model are constructed to recognize the four contact patterns based on the same dataset as those for the CNN-LSTM model. The average accuracies of the four contact patterns based on the CNN-LSTM, the CNN, and the RF algorithm are 99.43%, 96.67%, and 91.39%, respectively. All of the experimental results indicate that the CNN-LSTM constructed in this paper has very efficient performance in recognizing and classifying the contact patterns for the flexible tactile sensor.

Arrhythmias are defined as irregularities in the heartbeat rhythm, which may infrequently occur in a human’s life. These arrhythmias may cause potentially fatal complications, which may lead to an immediate risk of life. Thus, the detection and classification of arrhythmias is a pertinent issue for cardiac diagnosis. (1) Background: To capture these sporadic events, an electrocardiogram (ECG), a register containing the heart’s electrical function, is considered the gold standard. However, since ECG carries a vast amount of information, it becomes very complex and challenging to extract the relevant information from visual analysis. As a result, designing an efficient (automated) system to analyse the enormous quantity of data possessed by ECG is critical. (2) Method: This paper proposes a hybrid deep learning-based approach to automate the detection and classification process. This paper makes two-fold contributions. First, 1D ECG signals are translated into 2D Scalogram images to automate the noise filtering and feature extraction. Then, based on experimental evidence, by combining two learning models, namely 2D convolutional neural network (CNN) and the Long Short-Term Memory (LSTM) network, a hybrid model called 2D-CNN-LSTM is proposed. (3) Result: To evaluate the efficacy of the proposed 2D-CNN-LSTM approach, we conducted a rigorous experimental study using the widely adopted MIT–BIH arrhythmia database. The obtained results show that the proposed approach provides ≈98.7%, 99%, and 99% accuracy for Cardiac Arrhythmias (ARR), Congestive Heart Failure (CHF), and Normal Sinus Rhythm (NSR), respectively. Moreover, it provides an average sensitivity of the proposed model of 98.33% and a specificity value of 98.35%, for all three arrhythmias. (4) Conclusions: For the classification of arrhythmias, a robust approach has been introduced where 2D scalogram images of ECG signals are trained over the CNN-LSTM model. The results obtained are better as compared to the other existing techniques and will greatly reduce the amount of intervention required by doctors. For future work, the proposed method can be applied over some live ECG signals and Bi-LSTM can be applied instead of LSTM.

Pipeline transportation is the main method for long-distance gas transportation; however, ponding in the pipeline can affect transportation efficiency and even cause corrosion to the pipeline in some cases. A non-destructive method to detect pipeline ponding using percussion acoustic signals and a convolution neural network (CNN) is proposed in this paper. During the process of detection, a constant energy spring impact hammer is used to apply an impact on the pipeline, and the percussive acoustic signals are collected. A Mel spectrogram is used to extract the acoustic feature of the percussive acoustic signal with different ponding volumes in the pipeline. The Mel spectrogram is transferred to the input layer of the CNN and the convolutional kernel matrix of the CNN realizes the recognition of pipeline ponding volume. The recognition results show that the CNN can identify the amount of pipeline ponding with the percussive acoustic signals, which use the Mel spectrogram as the acoustic feature. Compared with the support vector machine (SVM) model and the decision tree model, the CNN model has better recognition performance. Therefore, the percussion-based pipeline ponding detection using the convolutional neural network method proposed in this paper has high application potential.

Based on experimental observations, there is a correlation between time and consecutive gaze positions in visual behaviors. Previous studies on gaze point estimation usually use images as the input for model trainings without taking into account the sequence relationship between image data. In addition to the spatial features, the temporal features are considered to improve the accuracy in this paper by using videos instead of images as the input data. To be able to capture spatial and temporal features at the same time, the convolutional neural network (CNN) and long short-term memory (LSTM) network are introduced to build a training model. In this way, CNN is used to extract the spatial features, and LSTM correlates temporal features. This paper presents a CNN Concatenating LSTM network (CCLN) that concatenates spatial and temporal features to improve the performance of gaze estimation in the case of time-series videos as the input training data. In addition, the proposed model can be optimized by exploring the numbers of LSTM layers, the influence of batch normalization (BN) and global average pooling layer (GAP) on CCLN. It is generally believed that larger amounts of training data will lead to better models. To provide data for training and prediction, we propose a method for constructing datasets of video for gaze point estimation. The issues are studied, including the effectiveness of different commonly used general models and the impact of transfer learning. Through exhaustive evaluation, it has been proved that the proposed method achieves a better prediction accuracy than the existing CNN-based methods. Finally, 93.1% of the best model and 92.6% of the general model MobileNet are obtained.

Radar signal intra-pulse modulation recognition can be addressed with convolutional neural networks (CNNs) and time–frequency images (TFIs). However, current CNNs have high computational complexity and do not perform well in low-signal-to-noise ratio (SNR) scenarios. In this paper, we propose a lightweight CNN known as the cross-scale aware network (CSANet) to recognize intra-pulse modulation based on three types of TFIs. The cross-scale aware (CSA) module, designed as a residual and parallel architecture, comprises a depthwise dilated convolution group (DDConv Group), a cross-channel interaction (CCI) mechanism, and spatial information focus (SIF). DDConv Group produces multiple-scale features with a dynamic receptive field, CCI fuses the features and mitigates noise in multiple channels, and SIF is aware of the cross-scale details of TFI structures. Furthermore, we develop a novel time–frequency fusion (TFF) feature based on three types of TFIs by employing image preprocessing techniques, i.e., adaptive binarization, morphological processing, and feature fusion. Experiments demonstrate that CSANet achieves higher accuracy with our TFF compared to other TFIs. Meanwhile, CSANet outperforms cutting-edge networks across twelve radar signal datasets, providing an efficient solution for high-precision recognition in low-SNR scenarios.

Sign language is undoubtedly a common way of communication among deaf and non-verbal people. But it is not common among hearing people to use sign language to express feelings or share information in everyday life. Therefore, a significant communication gap exists between deaf and hearing individuals, despite both groups experiencing similar emotions and sentiments. In this paper, we developed a convolutional neural network–squeeze excitation network to predict the sign language signs and developed a smartphone application to provide access to the ML model to use it. The SE block provides attention to the channel of the image, thus improving the performance of the model. On the other hand, the smartphone application brings the ML model close to people so that everyone can benefit from it. In addition, we used the Shapley additive explanation to interpret the black box nature of the ML model and understand the models working from within. Using our ML model, we achieved an accuracy of 99.86% on the KU-BdSL dataset. The SHAP analysis shows that the model primarily relies on hand-related visual cues to predict sign language signs, aligning with human communication patterns.

Intrapulse modulation classification of radar signals plays an important role in modern electronic reconnaissance, countermeasures, etc. In this paper, to improve the recognition rate at low signal-to-noise ratio (SNR), we propose a recognition method using the second-order short-time Fourier transform (STFT)-based synchrosqueezing transform (FSST2) combined with a modified convolution neural network, which we name MeNet. In particular, the radar signals are first preprocessed via the time–frequency analysis and STFT-based FSST2. Then, the informative features of the time–frequency images (TFIs) are deeply learned and classified through the MeNet with several specific convolutional blocks. The simulation results show that the overall recognition rate for seven types of intrapulse modulation radar signals can reach 95.6%, even when the SNR is −12 dB. Compared with other networks, the excellent recognition rate proves the superiority of our method.

During underwater image processing, image quality is affected by the absorption and scattering of light in water, thus causing problems such as blurring and noise. As a result, poor image quality is unavoidable. To achieve overall satisfying research results, underwater image denoising is vital. This paper presents an underwater image denoising method, named HHDNet, designed to address noise issues arising from environmental interference and technical limitations during underwater robot photography. The method leverages a dual-branch network architecture to handle both high and low frequencies, incorporating a hybrid attention module specifically designed for the removal of high-frequency abrupt noise in underwater images. Input images are decomposed into high-frequency and low-frequency components using a Gaussian kernel. For the high-frequency part, a Global Context Extractor (GCE) module with a hybrid attention mechanism focuses on removing high-frequency abrupt signals by capturing local details and global dependencies simultaneously. For the low-frequency part, efficient residual convolutional units are used in consideration of less noise information. Experimental results demonstrate that HHDNet effectively achieves underwater image denoising tasks, surpassing other existing methods not only in denoising effectiveness but also in maintaining computational efficiency, and thus HHDNet provides more flexibility in underwater image noise removal.

In the rapidly evolving fields of artificial intelligence and various industries, the secure processing and management of massive data have become paramount. This paper introduces an innovative reversible data hiding (RDH) method that leverages a Convolutional Neural Network (CNN)-based predictor to generate a predicted image from a given cover image. The secret data are ingeniously embedded within the differences in pixel values between the cover and predicted images. Our experimental analysis reveals a notable reduction in image distortion with increasing secret data size, showcasing the method’s potential for diverse applications. The unique aspect of our approach lies in the proportional relation between the Peak Signal-to-Noise Ratio (PSNR) and Embedding Capacity, highlighting its efficacy and efficiency in reversible data hiding.

Rapid and accurate grape leaf disease diagnosis is of great significance to its yield and quality of grape. In this paper, aiming at the identification of grape leaf diseases, a fast and accurate detection method based on fused deep features, extracted from a convolutional neural network (CNN), plus a support vector machine (SVM) is proposed. In the research, based on an open dataset, three types of state-of-the-art CNN networks, three kinds of deep feature fusion methods, seven species of deep feature layers, and a multi-class SVM classifier were studied. Firstly, images were resized to meet the input requirements of the CNN network; then, the deep features of the input images were extracted via the specific deep feature layer of the CNN network. Two kinds of deep features from different networks were then fused using different fusion methods to increase the effective classification feature information. Finally, a multi-class SVM classifier was trained with the fused deep features. The experimental results on the open dataset show that the fused deep features with any kind of fusion method can obtain a better classification performance than using a single type of deep feature. The direct concatenation of the Fc1000 deep feature extracted from ResNet50 and ResNet101 can achieve the best classification result compared with the other two fusion methods, and its F1 score is 99.81%. Furthermore, the SVM classifier trained using the proposed method can achieve a classification performance comparable to that of using the CNN model directly, but the training time is less than 1 s, which has an advantage over spending tens of minutes training a CNN model. The experimental results indicate that the method proposed in this paper can achieve fast and accurate identification of grape leaf diseases and meet the needs of actual agricultural production.

Phishing detection with high-performance accuracy and low computational complexity has always been a topic of great interest. New technologies have been developed to improve the phishing detection rate and reduce computational constraints in recent years. However, one solution is insufficient to address all problems caused by attackers in cyberspace. Therefore, the primary objective of this paper is to analyze the performance of various deep learning algorithms in detecting phishing activities. This analysis will help organizations or individuals select and adopt the proper solution according to their technological needs and specific applications’ requirements to fight against phishing attacks. In this regard, an empirical study was conducted using four different deep learning algorithms, including deep neural network (DNN), convolutional neural network (CNN), Long Short-Term Memory (LSTM), and gated recurrent unit (GRU). To analyze the behaviors of these deep learning architectures, extensive experiments were carried out to examine the impact of parameter tuning on the performance accuracy of the deep learning models. In addition, various performance metrics were measured to evaluate the effectiveness and feasibility of DL models in detecting phishing activities. The results obtained from the experiments showed that no single DL algorithm achieved the best measures across all performance metrics. The empirical findings from this paper also manifest several issues and suggest future research directions related to deep learning in the phishing detection domain.

Proper identification of different grape varieties by smart machinery is of great importance to modern agriculture production. In this paper, a fast and accurate identification method based on Canonical Correlation Analysis (CCA), which can fuse different deep features extracted from Convolutional Neural Network (CNN), plus Support Vector Machine (SVM) is proposed. In this research, based on an open dataset, three types of state-of-the-art CNNs, seven species of deep features, and a multi-class SVM classifier were studied. First, the images were resized to meet the input requirements of a CNN. Then, the deep features of the input images were extracted by a specific deep features layer of the CNN. Next, two kinds of deep features from different networks were fused by CCA to increase the effective classification feature information. Finally, a multi-class SVM classifier was trained with the fused features. When applied to an open dataset, the model outcome shows that the fused deep features with any combination can obtain better identification performance than by using a single type of deep feature. The fusion of fc6 (in AlexNet network) and Fc1000 (in ResNet50 network) deep features obtained the best identification performance. The average F1 Score of 96.9% was 8.7% higher compared to the best performance of a single deep feature, i.e., Fc1000 of ResNet101, which was 88.2%. Furthermore, the F1 Score of the proposed method is 2.7% higher than the best performance obtained by using a CNN directly. The experimental results show that the method proposed in this paper can achieve fast and accurate identification of grape varieties. Based on the proposed algorithm, the smart machinery in agriculture can take more targeted measures based on the different characteristics of different grape varieties for further improvement of the yield and quality of grape production.

Electric vehicles can reduce the dependence on limited resources such as oil, which is conducive to the development of clean energy. An accurate battery state of health (SOH) is beneficial for the safety of electric vehicles. A multi-feature and Convolutional Neural Network–Bidirectional Long Short-Term Memory–Multi-head Attention (CNN-BiLSTM-MHA)-based lithium-ion battery SOH estimation method is proposed in this paper. First, the voltage, energy, and temperature data of the battery in the constant current charging phase are measured. Then, based on the voltage and energy data, the incremental energy analysis (IEA) is performed to calculate the incremental energy (IE) curve. The IE curve features including IE, peak value, average value, and standard deviation are extracted and combined with the thermal features of the battery to form a complete multi-feature sequence. A CNN-BiLSTM-MHA model is set up to map the features to the battery SOH. Experiments were conducted using batteries with different charging currents, and the results showed that even if the nonlinearity of battery SOH degradation is significant, this method can still achieve a fast and accurate estimation of the battery SOH. The Mean Absolute Error (MAE) is 0.1982%, 0.1873%, 0.1652%, and 0.1968%, and the Root-Mean-Square Error (RMSE) is 0.2921%, 0.2997%, 0.2130%, and 0.2625%, respectively. The average Coefficient of Determination (R2) is above 96%. Compared to the BiLSTM model, the training time is reduced by an average of about 36%.

This paper proposes a convolutional neural network (CNN) model of the signal distribution control algorithm (SDCA) to maximize the dynamic vehicular traffic signal flow for each junction phase. The aim of the proposed algorithm is to determine the reward value and new state. It deconstructs the routing components of the current multi-directional queuing system (MDQS) architecture to identify optimal policies for every traffic scenario. Initially, the state value is divided into a function value and a parameter value. Combining these two scenarios updates the resulting optimized state value. Ultimately, an analogous criterion is developed for the current dataset. Next, the error or loss value for the present scenario is computed. Furthermore, utilizing the Deep Q-learning methodology with a quad agent enhances previous study discoveries. The recommended method outperforms all other traditional approaches in effectively optimizing traffic signal timing.

Recent advancements in 3D modeling have revolutionized various fields, including virtual reality, computer-aided diagnosis, and architectural design, emphasizing the importance of accurate quality assessment for 3D point clouds. As these models undergo operations such as simplification and compression, introducing distortions can significantly impact their visual quality. There is a growing need for reliable and efficient objective quality evaluation methods to address this challenge. In this context, this paper introduces a novel methodology to assess the quality of 3D point clouds using a deep learning-based no-reference (NR) method. First, it extracts geometric and perceptual attributes from distorted point clouds and represent them as a set of 1D vectors. Then, transfer learning is applied to obtain high-level features using a 1D convolutional neural network (1D CNN) adapted from 2D CNN models through weight conversion from ImageNet. Finally, quality scores are predicted through regression utilizing fully connected layers. The effectiveness of the proposed approach is evaluated across diverse datasets, including the Colored Point Cloud Quality Assessment Database (SJTU_PCQA), the Waterloo Point Cloud Assessment Database (WPC), and the Colored Point Cloud Quality Assessment Database featured at ICIP2020. The outcomes reveal superior performance compared to several competing methodologies, as evidenced by enhanced correlation with average opinion scores.

In recent years, deep learning methods have achieved remarkable success in hyperspectral image classification (HSIC), and the utilization of convolutional neural networks (CNNs) has proven to be highly effective. However, there are still several critical issues that need to be addressed in the HSIC task, such as the lack of labeled training samples, which constrains the classification accuracy and generalization ability of CNNs. To address this problem, a deep multi-scale attention fusion network (DMAF-NET) is proposed in this paper. This network is based on multi-scale features and fully exploits the deep features of samples from multiple levels and different perspectives with an aim to enhance HSIC results using limited samples. The innovation of this article is mainly reflected in three aspects: Firstly, a novel baseline network for multi-scale feature extraction is designed with a pyramid structure and densely connected 3D octave convolutional network enabling the extraction of deep-level information from features at different granularities. Secondly, a multi-scale spatial–spectral attention module and a pyramidal multi-scale channel attention module are designed, respectively. This allows modeling of the comprehensive dependencies of coordinates and directions, local and global, in four dimensions. Finally, a multi-attention fusion module is designed to effectively combine feature mappings extracted from multiple branches. Extensive experiments on four popular datasets demonstrate that the proposed method can achieve high classification accuracy even with fewer labeled samples.

The world’s expanding population presents a challenge through its rising demand for wood products. This requirement contributes to increased production and, ultimately, the high-quality and efficient utilization of basic materials. Detecting defects in wood elements, which are inevitable when working with a natural material such as wood, is one of the difficulties associated with the issue above. Even in modern times, people still identify wood defects by visually scrutinizing the sawn surface and marking the defects. Industrial scanners equipped with software based on convolutional neural networks (CNNs) allow for the rapid detection of defects and have the potential to accelerate production and eradicate human subjectivity. This paper evaluates the suitability of defect recognition software in industrial scanners against software specifically designed for this task within a research project conducted using Adaptive Vision Studio, focusing on feature detection techniques. The research revealed that the software installed as part of the industrial scanner is more effective for analyzing knots (77.78% vs. 70.37%), sapwood (100% vs. 80%), and ambrosia wood (60% vs. 20%), while the software derived from the project is more effective for analyzing cracks (70% vs. 65%), ingrown bark (42.86% vs. 28.57%), and wood rays (81.82% vs. 27.27%).

In this paper, a deep learning-based successive interference cancellation (SIC) scheme for use in nonorthogonal multiple access (NOMA) communication systems is investigated. NOMA has become a notable technique in the field of mobile wireless communication because of its capacity to overcome orthogonality, unlike a conventional orthogonal frequency division multiple access (OFDMA) communication system. In NOMA communication systems, SIC is one of the decoding schemes applied at receivers for downlink NOMA transmissions. In this paper, a convolutional neural network (CNN)-based SIC scheme is proposed to improve performance of the single base station and multiuser NOMA scheme. In contrast to existing SIC schemes, the proposed CNN-based SIC scheme can effectively mitigate losses resulting from imperfections of the SIC. The simulation results indicate that the CNN-based SIC method can successfully relieve conventional SIC impairments and achieve good detection performance. Consequently, a CNN-based SIC scheme can be considered as a potential technique for use in NOMA detection schemes.

This paper presents a machine learning approach to automatically classifying post-harvest vegetal species. Color images of vegetal species were applied to convolutional neural networks (CNNs) and support vector machine (SVM) classifiers. We focused on okra as the target vegetal species and classified it into two quality types. However, our approach could also be applied to other species. The machine learning solution consists of several components, and each design process and its combinations are essential for classification quality. Therefore, we carefully investigated their effects on classification accuracy. Through our experimental evaluation, we confirmed the following: (1) in color space selection, HLG (hue, lightness, and green) and HSL (hue, saturation, and lightness) are essential for vegetal species; (2) suitable preprocessing techniques are required owing to the complexity of the data and noise load; and (3) the diversity extension of learning image data by mixing different datasets obtained under different conditions is quite effective in reducing the overfitting possibility. The results of this study will assist AI practitioners in the design and development of post-harvest classifications based on machine learning.

Since Synthetic Aperture Radar (SAR) targets are full of coherent speckle noise, the traditional deep learning models are difficult to effectively extract key features of the targets and share high computational complexity. To solve the problem, an effective lightweight Convolutional Neural Network (CNN) model incorporating transfer learning is proposed for better handling SAR targets recognition tasks. In this work, firstly we propose the Atrous-Inception module, which combines both atrous convolution and inception module to obtain rich global receptive fields, while strictly controlling the parameter amount and realizing lightweight network architecture. Secondly, the transfer learning strategy is used to effectively transfer the prior knowledge of the optical, non-optical, hybrid optical and non-optical domains to the SAR target recognition tasks, thereby improving the model’s recognition performance on small sample SAR target datasets. Finally, the model constructed in this paper is verified to be 97.97% on ten types of MSTAR datasets under standard operating conditions, reaching a mainstream target recognition rate. Meanwhile, the method presented in this paper shows strong robustness and generalization performance on a small number of randomly sampled SAR target datasets.

This paper aims to explore and evaluate aerial imagery and deep learning technology in pavement condition evaluation. A convolutional neural network (CNN) model, named PCIer, was designed to process aerial images and produce pavement condition index (PCI) estimations, which are classified into four scales of Good (PCI ≥ 70), Fair (50 ≤ PCI < 70), Poor (25 ≤ PCI < 50), and Very Poor (PCI < 25). In the experiment, the PCI datasets were retrieved from the published pavement condition report by the City of Sacramento, CA. Following the retrieved datasets, the authors also collected the corresponding aerial image datasets containing 100 images for each PCI grade from Google Earth. An 80% proportion of datasets were used for PCIer model training, and the remaining were used for testing. Comparisons showed using a 128-channel heatmap layer in the proposed PCIer model and saving the PCIer model with the best validation accuracy would yield the best performance, with a testing accuracy of 0.97, and a weighted average precision, recall, and F1-score of 0.98, 0.97, and 0.97, respectively. Moreover, future research recommendations are provided in the discussion for improving the effectiveness of pavement evaluation via aerial imagery and deep learning.

The monitoring of the lifetime of cutting tools often faces problems such as life data loss, drift, and distortion. The prediction of the lifetime in this situation is greatly compromised with respect to the accuracy. The recent rise of deep learning, such as Gated Recurrent Unit Units (GRUs), Hidden Markov Models (HMMs), Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), Attention networks, and Transformers, has dramatically improved the data problems in tool lifetime prediction, substantially enhancing the accuracy of tool wear prediction. In this paper, we introduce a novel approach known as PCHIP-Enhanced ConvGRU (PECG), which leverages multiple—feature fusion for tool wear prediction. When compared to traditional models such as CNNs, the CNN Block, and GRUs, our method consistently outperformed them across all key performance metrics, with a primary focus on the accuracy. PECG addresses the challenge of missing tool wear measurement data in relation to sensor data. By employing PCHIP interpolation to fill in the gaps in the wear values, we have developed a model that combines the strengths of both CNNs and GRUs with data augmentation. The experimental results demonstrate that our proposed method achieved an exceptional relative accuracy of 0.8522, while also exhibiting a Pearson’s Correlation Coefficient (PCC) exceeding 0.95. This innovative approach not only predicts tool wear with remarkable precision, but also offers enhanced stability.

Deep learning (DL) networks are a popular choice for short-term load forecasting (STLF) in the residential sector. Hybrid DL methodologies based on convolutional neural networks (CNNs) and long short-term memory networks (LSTMs) have a higher forecasting accuracy than conventional statistical STLF techniques for different types of single-household load series. However, existing load forecasting methodologies are often inefficient when a high load demand persists for a few hours in a day. Peak load consumption is explicitly depicted as a tail in the probability distribution function (PDF) of the load series. Due to the diverse and uncertain nature of peak load demands, DL methodologies have difficulty maintaining consistent forecasting results, particularly when the PDF of the load series has a longer tail. This paper proposes a multihousehold load forecasting strategy based on the collective moment measure (CMM) (which is obtained from the PDF of the load series), data augmentation, and a CNN. Each load series was compared and ordered through CMM indexing, which helped maintain a minimum or constant shifting variance in the dataset inputted to the CNN. Data augmentation was used to enlarge the input dataset and solve the existing data requirement issues of the CNN. With the ordered load series and data augmentation strategy, the simulation results demonstrated a significant improvement in the performance of both single-household and multihousehold load forecasting. The proposed method predicts day-ahead multihousehold loads simultaneously and compares the results based on a single household. The forecasting performance of the proposed method for six different household groups with 10, 20, 30, 50, 80, and 100 household load series was evaluated and compared with those of existing methodologies. The mean absolute percentage error of the prediction results for each multihousehold load series could be improved by more than 3%. This study can help advance the application of DL methods for household load prediction under high-load-demand conditions.

Flexible pressure sensors play a crucial role in detecting human motion and facilitating human–computer interaction. In this paper, a type of flexible pressure sensor unit with high sensitivity (2.242 kPa−1), fast response time (80 ms), and remarkable stability (1000 cycles) is proposed and fabricated by the multi-walled carbon nanotube (MWCNT)/cotton fabric (CF) material based on a dip-coating method. Six flexible pressure sensor units are integrated into a flexible wristband and made into a wearable and portable wrist sensor with favorable stability. Then, seven wrist gestures (Gesture Group #1), five letter gestures (Gesture Group #2), and eight sign language gestures (Gesture Group #3) are performed by wearing the wrist sensor, and the corresponding time sequence signals of the three gesture groups (#1, #2, and #3) from the wrist sensor are collected, respectively. To efficiently recognize different gestures from the three groups detected by the wrist sensor, a fusion network model combined with a convolutional neural network (CNN) and the bidirectional long short-term memory (BiLSTM) neural network, named CNN-BiLSTM, which has strong robustness and generalization ability, is constructed. The three types of Gesture Groups were recognized based on the CNN-BiLSTM model with accuracies of 99.40%, 95.00%, and 98.44%. Twenty gestures (merged by Group #1, #2, and #3) were recognized with an accuracy of 96.88% to validate the applicability of the wrist sensor based on this model for gesture recognition. The experimental results denote that the CNN-BiLSTM model has very efficient performance in recognizing different gestures collected from the flexible wrist sensor.

In recent years, radar emitter signal recognition has enjoyed a wide range of applications in electronic support measure systems and communication security. More and more deep learning algorithms have been used to improve the recognition accuracy of radar emitter signals. However, complex deep learning algorithms and data preprocessing operations have a huge demand for computing power, which cannot meet the requirements of low power consumption and high real-time processing scenarios. Therefore, many research works have remained in the experimental stage and cannot be actually implemented. To tackle this problem, this paper proposes a resource reuse computing acceleration platform based on field programmable gate arrays (FPGA), and implements a one-dimensional (1D) convolutional neural network (CNN) and long short-term memory (LSTM) neural network (NN) model for radar emitter signal recognition, directly targeting the intermediate frequency (IF) data of radar emitter signal for classification and recognition. The implementation of the 1D-CNN-LSTM neural network on FPGA is realized by multiplexing the same systolic array to accomplish the parallel acceleration of 1D convolution and matrix vector multiplication operations. We implemented our network on Xilinx XCKU040 to evaluate the effectiveness of our proposed solution. Our experiments show that the system can achieve 7.34 giga operations per second (GOPS) data throughput with only 5.022 W power consumption when the radar emitter signal recognition rate is 96.53%, which greatly improves the energy efficiency ratio and real-time performance of the radar emitter recognition system.

Forest stock volume is the main factor to evaluate forest carbon sink level. At present, the combination of multi-source remote sensing and non-parametric models has been widely used in FSV estimation. However, the biodiversity of natural forests is complex, and the response of the spatial information of remote sensing images to FSV is significantly reduced, which seriously affects the accuracy of FSV estimation. To address this challenge, this paper takes China’s Baishanzu Forest Park with representative characteristics of natural forests as the research object, integrates the forest survey data, SRTM data, and Landsat 8 images of Baishanzu Forest Park, constructs a time series dataset based on survey time, and establishes an FSV estimation model based on the CNN-LSTM-Attention algorithm. The model uses the convolutional neural network to extract the spatial features of remote sensing images, uses the LSTM to capture the time-varying characteristics of FSV, captures the feature variables with a high response to FSV through the attention mechanism, and finally completes the prediction of FSV. The experimental results show that some features (e.g., texture, elevation, etc.) of the dataset based on multi-source data feature variables are more effective in FSV estimation than spectral features. Compared with the existing models such as MLR and RF, the proposed model achieved higher accuracy in the study area (R2 = 0.8463, rMSE = 26.73 m3/ha, MAE = 16.47 m3/ha).

This paper proposed a real-time fault diagnostic method for hydraulic systems using data collected from multiple sensors. The method is based on a proposed multi-sensor convolutional neural network (MS-CNN) that incorporates feature extraction, sensor selection, and fault diagnosis into an end-to-end model. Both the sensor selection process and fault diagnosis process are based on abstract fault-related features learned by a CNN deep learning model. Therefore, compared with the traditional sensor-and-feature selection method, the proposed MS-CNN can find the sensor channels containing higher-level fault-related features, which provides two advantages for diagnosis. First, the sensor selection can reduce the redundant information and improve the diagnostic performance of the model. Secondly, the reduced number of sensors simplifies the model, reducing communication burden and computational complexity. These two advantages make the MS-CNN suitable for real-time hydraulic system fault diagnosis, in which the multi-sensor feature extraction and the computation speed are both significant. The proposed MS-CNN approach is evaluated experimentally on an electric-hydraulic subsea control system test rig and an open-source dataset. The proposed method shows obvious superiority in terms of both diagnosis accuracy and computational speed when compared with traditional CNN models and other state-of-the-art multi-sensor diagnostic methods.

The idea of Ubiquitous Power Internet of Things (UPIoTs) accelerates the development of intelligent monitoring and diagnostic technologies. In this paper, a diagnostic method suitable for power equipment in an interference environment was proposed based on the deep Convolutional Neural Network (CNN): MobileNet-V2 and Digital Image Processing (DIP) methods to conduct fault identification process: including fault type classification and fault localization. A data visualization theory was put forward in this paper, which was applied in frequency response (FR) curves of transformer to obtain dataset. After the image augmentation process, the dataset was input into the deep CNN: MobileNet-V2 for training procedures. Then a spatial-probabilistic mapping relationship was established based on traditional Frequency Response Analysis (FRA) fault diagnostic method. Each image in the dataset was compared with the fingerprint values to get traditional diagnosing results. Next, the anti-interference abilities of the proposed CNN-DIP method were compared with that of the traditional one while the magnitude of the interference gradually increased. Finally, the fault tolerance of the proposed method was verified by further analyzing the deviations between the wrong diagnosing results with the corresponding actual labels. Experimental results showed that the proposed deep visual identification (CNN-DIP) method has a higher diagnosing accuracy, a stronger anti-interference ability and a better fault tolerance.

Global Positioning System (GPS) has been used in many aerial and terrestrial high precision positioning applications. Multipath affects positioning and navigation performance. This paper proposes a convolutional neural network based carrier-phase multipath detection method. The method is based on the fact that the features of multipath characteristics in multipath contaminated data can be learned and identified by a convolutional neural network. The proposed method is validated with simulated and real GPS data and compared with existing multipath mitigation methods in position domain. The results show the proposed method can detect about 80% multipath errors (i.e., recall) in both simulated and real data. The impact of the proposed method on positioning accuracy improvement is demonstrated with two datasets, 18⁻30% improvement is obtained by down-weighting the detected multipath measurements. The focus of this paper is on the development and test of the proposed convolutional neural network based multipath detection algorithm.

The automatic monitoring of activities in construction sites through the proper use of acoustic signals is a recent field of research that is currently in continuous evolution. In particular, the use of techniques based on Convolutional Neural Networks (CNNs) working on the spectrogram of the signal or its mel-scale variants was demonstrated to be quite successful. Nevertheless, the spectrogram has some limitations, which are due to the intrinsic trade-off between temporal and spectral resolutions. In order to overcome these limitations, in this paper, we propose employing the scalogramas a proper time–frequency representation of the audio signal. The scalogram is defined as the square modulus of the Continuous Wavelet Transform (CWT) and is known as a powerful tool for analyzing real-world signals. Experimental results, obtained on real-world sounds recorded in construction sites, have demonstrated the effectiveness of the proposed approach, which is able to clearly outperform most state-of-the-art solutions.

A Brain–Computer Interface (BCI) is a medium for communication between the human brain and computers, which does not rely on other human neural tissues, but only decodes Electroencephalography (EEG) signals and converts them into commands to control external devices. Motor Imagery (MI) is an important BCI paradigm that generates a spontaneous EEG signal without external stimulation by imagining limb movements to strengthen the brain’s compensatory function, and it has a promising future in the field of computer-aided diagnosis and rehabilitation technology for brain diseases. However, there are a series of technical difficulties in the research of motor imagery-based brain–computer interface (MI-BCI) systems, such as: large individual differences in subjects and poor performance of the cross-subject classification model; a low signal-to-noise ratio of EEG signals and poor classification accuracy; and the poor online performance of the MI-BCI system. To address the above problems, this paper proposed a combined virtual electrode-based EEG Source Analysis (ESA) and Convolutional Neural Network (CNN) method for MI-EEG signal feature extraction and classification. The outcomes reveal that the online MI-BCI system developed based on this method can improve the decoding ability of multi-task MI-EEG after training, it can learn generalized features from multiple subjects in cross-subject experiments and has some adaptability to the individual differences of new subjects, and it can decode the EEG intent online and realize the brain control function of the intelligent cart, which provides a new idea for the research of an online MI-BCI system.

The tunnel construction area poses significant challenges for the use of vision technology due to the presence of nonhomogeneous haze fields and low-contrast targets. However, existing dehazing algorithms display weak generalization, leading to dehazing failures, incomplete dehazing, or color distortion in this scenario. Therefore, an adversarial dual-branch convolutional neural network (ADN) is proposed in this paper to deal with the above challenges. The ADN utilizes two branches of the knowledge transfer sub-network and the multi-scale dense residual sub-network to process the hazy image and then aggregate the channels. This input is then passed through a discriminator to judge true and false, motivating the network to improve performance. Additionally, a tunnel haze field simulation dataset (Tunnel-HAZE) is established based on the characteristics of nonhomogeneous dust distribution and artificial light sources in the tunnel. Comparative experiments with existing advanced dehazing algorithms indicate an improvement in both PSNR (Peak Signal-to-Noise Ratio) and SSIM (Structural Similarity) by 4.07 dB and 0.032 dB, respectively. Furthermore, a binocular measurement experiment conducted in a simulated tunnel environment demonstrated a reduction in the relative error of measurement results by 50.5% when compared to the haze image. The results demonstrate the effectiveness and application potential of the proposed method in tunnel construction.