Search

To ensure the safe operation of highway traffic lines, given the imperfect feature extraction of existing road pit defect detection models and the practicability of detection equipment, this paper proposes a lightweight target detection algorithm with enhanced feature extraction based on the YOLO (You Only Look Once) algorithm. The BIFPN (Bidirectional Feature Pyramid Network) network structure is used for multi-scale feature fusion to enhance the feature extraction ability, and Varifocal Loss is used to optimize the sample imbalance problem, which improves the accuracy of road defect target detection. In the evaluation test of the model in the constructed PCD1 (Pavement Check Dataset) dataset, the mAP@.5 (mean Average Precision when IoU = 0.5) of the BV-YOLOv5S (BiFPN Varifocal Loss-YOLOv5S) model increased by 4.1%, 3%, and 0.9%, respectively, compared with the YOLOv3-tiny, YOLOv5S, and B-YOLOv5S (BiFPN-YOLOv5S; BV-YOLOv5S does not use the Improved Focal Loss function) models. Through the analysis and comparison of experimental results, it is proved that the proposed BV-YOLOv5S network model performs better and is more reliable in the detection of pavement defects and can meet the needs of road safety detection projects with high real-time and flexibility requirements. [ABSTRACT FROM AUTHOR]

Hand gesture recognition (HGR) is a crucial area of research that enhances communication by overcoming language barriers and facilitating human-computer interaction. Although previous works in HGR have employed deep neural networks, they fail to encode the orientation and position of the hand in the image. To address this issue, this paper proposes HGR-ViT, a Vision Transformer (ViT) model with an attention mechanism for hand gesture recognition. Given a hand gesture image, it is first split into fixed size patches. Positional embedding is added to these embeddings to form learnable vectors that capture the positional information of the hand patches. The resulting sequence of vectors are then served as the input to a standard Transformer encoder to obtain the hand gesture representation. A multilayer perceptron head is added to the output of the encoder to classify the hand gesture to the correct class. The proposed HGR-ViT obtains an accuracy of 99.98%, 99.36% and 99.85% for the American Sign Language (ASL) dataset, ASL with Digits dataset, and National University of Singapore (NUS) hand gesture dataset, respectively. [ABSTRACT FROM AUTHOR]

Deep unrolling networks (DUNs) have emerged as a promising approach for solving compressed sensing (CS) problems due to their superior explainability, speed, and performance compared to classical deep network models. However, the CS performance in terms of efficiency and accuracy remains a principal challenge for approaching further improvements. In this paper, we propose a novel deep unrolling model, SALSA-Net, to solve the image CS problem. The network architecture of SALSA-Net is inspired by unrolling and truncating the split augmented Lagrangian shrinkage algorithm (SALSA) which is used to solve sparsity-induced CS reconstruction problems. SALSA-Net inherits the interpretability of the SALSA algorithm while incorporating the learning ability and fast reconstruction speed of deep neural networks. By converting the SALSA algorithm into a deep network structure, SALSA-Net consists of a gradient update module, a threshold denoising module, and an auxiliary update module. All parameters, including the shrinkage thresholds and gradient steps, are optimized through end-to-end learning and are subject to forward constraints to ensure faster convergence. Furthermore, we introduce learned sampling to replace traditional sampling methods so that the sampling matrix can better preserve the feature information of the original signal and improve sampling efficiency. Experimental results demonstrate that SALSA-Net achieves significant reconstruction performance compared to state-of-the-art methods while inheriting the advantages of explainable recovery and high speed from the DUNs paradigm. [ABSTRACT FROM AUTHOR]

Human activity recognition (HAR) is an important research problem in computer vision. This problem is widely applied to building applications in human–machine interactions, monitoring, etc. Especially, HAR based on the human skeleton creates intuitive applications. Therefore, determining the current results of these studies is very important in selecting solutions and developing commercial products. In this paper, we perform a full survey on using deep learning to recognize human activity based on three-dimensional (3D) human skeleton data as input. Our research is based on four types of deep learning networks for activity recognition based on extracted feature vectors: Recurrent Neural Network (RNN) using extracted activity sequence features; Convolutional Neural Network (CNN) uses feature vectors extracted based on the projection of the skeleton into the image space; Graph Convolution Network (GCN) uses features extracted from the skeleton graph and the temporal–spatial function of the skeleton; Hybrid Deep Neural Network (Hybrid–DNN) uses many other types of features in combination. Our survey research is fully implemented from models, databases, metrics, and results from 2019 to March 2023, and they are presented in ascending order of time. In particular, we also carried out a comparative study on HAR based on a 3D human skeleton on the KLHA3D 102 and KLYOGA3D datasets. At the same time, we performed analysis and discussed the obtained results when applying CNN-based, GCN-based, and Hybrid–DNN-based deep learning networks. [ABSTRACT FROM AUTHOR]

Aiming at the problems of long detection time and low detection accuracy in the existing coal gangue recognition, this paper proposes a method to collect the multispectral images of coal gangue using spectral technology and match with the improved YOLOv5s (You Only Look Once Version-5s) neural network model to apply it to coal gangue target recognition and detection, which can effectively reduce the detection time and improve the detection accuracy and recognition effect of coal gangue. In order to take the coverage area, center point distance and aspect ratio into account at the same time, the improved YOLOv5s neural network replaces the original GIou Loss loss function with CIou Loss loss function. At the same time, DIou NMS replaces the original NMS, which can effectively detect overlapping targets and small targets. In the experiment, 490 sets of multispectral data were obtained through the multispectral data acquisition system. Using the random forest algorithm and the correlation analysis of bands, the spectral images of the sixth, twelfth and eighteenth bands from twenty-five bands were selected to form a pseudo RGB image. A total of 974 original sample images of coal and gangue were obtained. Through two image noise reduction methods, namely, Gaussian filtering algorithm and non-local average noise reduction, 1948 images of coal gangue were obtained after preprocessing the dataset. This was divided into a training set and test set according to an 8:2 ratio and trained in the original YOLOv5s neural network, improved YOLOv5s neural network and SSD neural network. By identifying and detecting the three neural network models obtained after training, the results can be obtained, the loss value of the improved YOLOv5s neural network model is smaller than the original YOLOv5s neural network and SSD neural network, the recall rate is closer to 1 than the original YOLOv5s neural network and SSD neural network, the detection time is the shortest, the recall rate is 100% and the average detection accuracy of coal and gangue is the highest. The average precision of the training set is increased to 0.995, which shows that the improved YOLOv5s neural network has a better effect on the detection and recognition of coal gangue. The detection accuracy of the improved YOLOv5s neural network model test set is increased from 0.73 to 0.98, and all overlapping targets can also be accurately detected without false detection or missed detection. At the same time, the size of the improved YOLOv5s neural network model after training is reduced by 0.8 MB, which is conducive to hardware transplantation. [ABSTRACT FROM AUTHOR]

In integrated circuit manufacturing, defects in epoxy drops for die attachments are required to be identified during production. Modern identification techniques based on vision-based deep neural networks require the availability of a very large number of defect and non-defect epoxy drop images. In practice, however, very few defective epoxy drop images are available. This paper presents a generative adversarial network solution to generate synthesized defective epoxy drop images as a data augmentation approach so that vision-based deep neural networks can be trained or tested using such images. More specifically, the so-called CycleGAN variation of the generative adversarial network is used by enhancing its cycle consistency loss function with two other loss functions consisting of learned perceptual image patch similarity (LPIPS) and a structural similarity index metric (SSIM). The results obtained indicate that when using the enhanced loss function, the quality of synthesized defective epoxy drop images are improved by 59%, 12%, and 131% for the metrics of the peak signal-to-noise ratio (PSNR), universal image quality index (UQI), and visual information fidelity (VIF), respectively, compared to the CycleGAN standard loss function. A typical image classifier is used to show the improvement in the identification outcome when using the synthesized images generated by the developed data augmentation approach. [ABSTRACT FROM AUTHOR]

In advanced transportation-management systems, variable speed limits are a crucial application. Deep reinforcement learning methods have been shown to have superior performance in many applications, as they are an effective approach to learning environment dynamics for decision-making and control. However, they face two significant difficulties in traffic-control applications: reward engineering with delayed reward and brittle convergence properties with gradient descent. To address these challenges, evolutionary strategies are well suited as a class of black-box optimization techniques inspired by natural evolution. Additionally, the traditional deep reinforcement learning framework struggles to handle the delayed reward setting. This paper proposes a novel approach using covariance matrix adaptation evolution strategy (CMA-ES), a gradient-free global optimization method, to handle the task of multi-lane differential variable speed limit control. The proposed method uses a deep-learning-based method to dynamically learn optimal and distinct speed limits among lanes. The parameters of the neural network are sampled using a multivariate normal distribution, and the dependencies between the variables are represented by a covariance matrix that is optimized dynamically by CMA-ES based on the freeway's throughput. The proposed approach is tested on a freeway with simulated recurrent bottlenecks, and the experimental results show that it outperforms deep reinforcement learning-based approaches, traditional evolutionary search methods, and the no-control scenario. Our proposed method demonstrates a 23% improvement in average travel time and an average of a 4% improvement in CO, HC, and NOx emission.Furthermore, the proposed method produces explainable speed limits and has desirable generalization power. [ABSTRACT FROM AUTHOR]

The paper was devoted to the application of saliency analysis methods in the performance analysis of deep neural networks used for the binary classification of brain tumours. We have presented the basic issues related to deep learning techniques. A significant challenge in using deep learning methods is the ability to explain the decision-making process of the network. To ensure accurate results, the deep network being used must undergo extensive training to produce high-quality predictions. There are various network architectures that differ in their properties and number of parameters. Consequently, an intriguing question is how these different networks arrive at similar or distinct decisions based on the same set of prerequisites. Therefore, three widely used deep convolutional networks have been discussed, such as VGG16, ResNet50 and EfficientNetB7, which were used as backbone models. We have customized the output layer of these pre-trained models with a softmax layer. In addition, an additional network has been described that was used to assess the saliency areas obtained. For each of the above networks, many tests have been performed using key metrics, including statistical evaluation of the impact of class activation mapping (CAM) and gradient-weighted class activation mapping (Grad-CAM) on network performance on a publicly available dataset of brain tumour X-ray images. [ABSTRACT FROM AUTHOR]

In recent years, convolutional neural networks have been in the leading position for ground-based cloud image classification tasks. However, this approach introduces too much inductive bias, fails to perform global modeling, and gradually tends to saturate the performance effect of convolutional neural network models as the amount of data increases. In this paper, we propose a novel method for ground-based cloud image recognition based on the multi-modal Swin Transformer (MMST), which discards the idea of using convolution to extract visual features and mainly consists of an attention mechanism module and linear layers. The Swin Transformer, the visual backbone network of MMST, enables the model to achieve better performance in downstream tasks through pre-trained weights obtained from the large-scale dataset ImageNet and can significantly shorten the transfer learning time. At the same time, the multi-modal information fusion network uses multiple linear layers and a residual structure to thoroughly learn multi-modal features, further improving the model's performance. MMST is evaluated on the multi-modal ground-based cloud public data set MGCD. Compared with the state-of-art methods, the classification accuracy rate reaches 91.30%, which verifies its validity in ground-based cloud image classification and proves that in ground-based cloud image recognition, models based on the Transformer architecture can also achieve better results. [ABSTRACT FROM AUTHOR]

Recently, many deep neural networks (DNN) have been proposed to solve the spectral reconstruction (SR) problem: recovering spectra from RGB measurements. Most DNNs seek to learn the relationship between an RGB viewed in a given spatial context and its corresponding spectra. Significantly, it is argued that the same RGB can map to different spectra depending on the context with respect to which it is seen and, more generally, that accounting for spatial context leads to improved SR. However, as it stands, DNN performance is only slightly better than the much simpler pixel-based methods where spatial context is not used. In this paper, we present a new pixel-based algorithm called A++ (an extension of the A+ sparse coding algorithm). In A+, RGBs are clustered, and within each cluster, a designated linear SR map is trained to recover spectra. In A++, we cluster the spectra instead in an attempt to ensure neighboring spectra (i.e., spectra in the same cluster) are recovered by the same SR map. A polynomial regression framework is developed to estimate the spectral neighborhoods given only the RGB values in testing, which in turn determines which mapping should be used to map each testing RGB to its reconstructed spectrum. Compared to the leading DNNs, not only does A++ deliver the best results, it is parameterized by orders of magnitude fewer parameters and has a significantly faster implementation. Moreover, in contradistinction to some DNN methods, A++ uses pixel-based processing, which is robust to image manipulations that alter the spatial context (e.g., blurring and rotations). Our demonstration on the scene relighting application also shows that, while SR methods, in general, provide more accurate relighting results compared to the traditional diagonal matrix correction, A++ provides superior color accuracy and robustness compared to the top DNN methods. [ABSTRACT FROM AUTHOR]

Collision-free trajectory planning in narrow spaces has become one of the most challenging tasks in automated parking scenarios. Previous optimization-based approaches can generate accurate parking trajectories, but these methods cannot compute feasible solutions with extremely complex constraints in a limited time. Recent research uses neural-network-based approaches that can generate time-optimized parking trajectories in linear time. However, the generalization of these neural network models in different parking scenarios has not been considered thoroughly and the risk of privacy compromise exists in the case of centralized training. To address the above issues, this paper proposes a hierarchical trajectory planning method with deep reinforcement learning in the federated learning scheme (HALOES) to rapidly and accurately generate collision-free automated parking trajectories in multiple narrow spaces. HALOES is a federated learning based hierarchical trajectory planning method to fully exert high-level deep reinforcement learning and the low-level optimization-based approach. HALOES further fuse the deep reinforcement learning model parameters to improve the generalization capabilities with a decentralized training scheme. The federated learning scheme in HALOES aims to protect the privacy of the vehicle's data during model parameter aggregation. Simulation results show that the proposed method can achieve efficient automatic parking in multiple narrow spaces, improve planning time from 12.15 % to 66.02 % compared to other state-of-the-art methods (e.g., hybrid A*, OBCA) and maintain the same level of trajectory accuracy while having great model generalization. [ABSTRACT FROM AUTHOR]

Regularization is an important technique for training deep neural networks. In this paper, we propose a novel shared-weight teacher–student strategy and a content-aware regularization (CAR) module. Based on a tiny, learnable, content-aware mask, CAR is randomly applied to some channels in the convolutional layers during training to be able to guide predictions in a shared-weight teacher–student strategy. CAR prevents motion estimation methods in unsupervised learning from co-adaptation. Extensive experiments on optical flow and scene flow estimation show that our method significantly improves on the performance of the original networks and surpasses other popular regularization methods. The method also surpasses all variants with similar architectures and the supervised PWC-Net on MPI-Sintel and on KITTI. Our method shows strong cross-dataset generalization, i.e., our method solely trained on MPI-Sintel outperforms a similarly trained supervised PWC-Net by 27.9% and 32.9% on KITTI, respectively. Our method uses fewer parameters and less computation, and has faster inference times than the original PWC-Net. [ABSTRACT FROM AUTHOR]

Semi-supervised learning is a learning pattern that can utilize labeled data and unlabeled data to train deep neural networks. In semi-supervised learning methods, self-training-based methods do not depend on a data augmentation strategy and have better generalization ability. However, their performance is limited by the accuracy of predicted pseudo-labels. In this paper, we propose to reduce the noise in the pseudo-labels from two aspects: the accuracy of predictions and the confidence of the predictions. For the first aspect, we propose a similarity graph structure learning (SGSL) model that considers the correlation between unlabeled and labeled samples, which facilitates the learning of more discriminative features and, thus, obtains more accurate predictions. For the second aspect, we propose an uncertainty-based graph convolutional network (UGCN), which can aggregate similar features based on the learned graph structure in the training phase, making the features more discriminative. It can also output the uncertainty of predictions in the pseudo-label generation phase, generating pseudo-labels only for unlabeled samples with low uncertainty; thus, reducing the noise in the pseudo-labels. Further, a positive and negative self-training framework is proposed, which combines the proposed SGSL model and UGCN into the self-training framework for end-to-end training. In addition, in order to introduce more supervised signals in the self-training process, negative pseudo-labels are generated for unlabeled samples with low prediction confidence, and then the positive and negative pseudo-labeled samples are trained together with a small number of labeled samples to improve the performance of semi-supervised learning. The code is available upon request. [ABSTRACT FROM AUTHOR]

The research presents preliminary results where the smartphone camera and ambient light sensor can be used to communicate with road infrastructure, vehicles, and pedestrians. The explosive growth and wide proliferation of mobile devices, the majority of which are smartphones, led to the inception of several novel and intuitive services, including on-the-go services, online customer services, and location-based services (LBS). With the excessive use of smartphones by the masses globally, more and more sensors are being embedded in smartphones. Micro-electro-mechanical-systems sensors, embedded in modern smartphones, are used to record the physical movements of users, temperature readings, and cough sounds with audio sensors. [Extracted from the article]

Microcontrollers (MCUs) have been deployed on numerous IoT devices due to their compact sizes and low costs. MCUs are capable of capturing sensor data and processing them. However, due to their low computational power, applications processing sensor data with deep neural networks (DNNs) have been limited. In this paper, we propose MiCrowd, a floating population measurement system with a tiny DNNs running on MCUs since the data have essential value in urban planning and business. Moreover, MiCrowd addresses the following important challenges: (1) privacy issues, (2) communication costs, and (3) extreme resource constraints on MCUs. To tackle those challenges, we designed a lightweight crowd-counting deep neural network, named MiCrowdNet, which enables on-MCU inferences. In addition, our dataset is carefully chosen and completely re-labeled to train MiCrowdNet for counting people from an mobility view. Experiments show the effectiveness of MiCrowdNet and our relabeled dataset for accurate on-device crowd counting. [ABSTRACT FROM AUTHOR]

The tremendous growth in online activity and the Internet of Things (IoT) led to an increase in cyberattacks. Malware infiltrated at least one device in almost every household. Various malware detection methods that use shallow or deep IoT techniques were discovered in recent years. Deep learning models with a visualization method are the most commonly and popularly used strategy in most works. This method has the benefit of automatically extracting features, requiring less technical expertise, and using fewer resources during data processing. Training deep learning models that generalize effectively without overfitting is not feasible or appropriate with large datasets and complex architectures. In this paper, a novel ensemble model, Stacked Ensemble—autoencoder, GRU, and MLP or SE-AGM, composed of three light-weight neural network models—autoencoder, GRU, and MLP—that is trained on the 25 essential and encoded extracted features of the benchmark MalImg dataset for classification was proposed. The GRU model was tested for its suitability in malware detection due to its lesser usage in this domain. The proposed model used a concise set of malware features for training and classifying the malware classes, which reduced the time and resource consumption in comparison to other existing models. The novelty lies in the stacked ensemble method where the output of one intermediate model works as input for the next model, thereby refining the features as compared to the general notion of an ensemble approach. Inspiration was drawn from earlier image-based malware detection works and transfer learning ideas. To extract features from the MalImg dataset, a CNN-based transfer learning model that was trained from scratch on domain data was used. Data augmentation was an important step in the image processing stage to investigate its effect on classifying grayscale malware images in the MalImg dataset. SE-AGM outperformed existing approaches on the benchmark MalImg dataset with an average accuracy of 99.43%, demonstrating that our method was on par with or even surpassed them. [ABSTRACT FROM AUTHOR]

Robotic systems are a fundamental part of modern industrial development. In this regard, they are required for long periods, in repetitive processes that must comply with strict tolerance ranges. Hence, the positional accuracy of the robots is critical, since degradation of this can represent a considerable loss of resources. In recent years, prognosis and health management (PHM) methodologies, based on machine and deep learning, have been applied to robots, in order to diagnose and detect faults and identify the degradation of robot positional accuracy, using external measurement systems, such as lasers and cameras; however, their implementation is complex in industrial environments. In this respect, this paper proposes a method based on discrete wavelet transform, nonlinear indices, principal component analysis, and artificial neural networks, in order to detect a positional deviation in robot joints, by analyzing the currents of the actuators. The results show that the proposed methodology allows classification of the robot positional degradation with an accuracy of 100%, using its current signals. The early detection of robot positional degradation, allows the implementation of PHM strategies on time, and prevents losses in manufacturing processes. [ABSTRACT FROM AUTHOR]

Spiking neural networks (SNNs) are subjects of a topic that is gaining more and more interest nowadays. They more closely resemble actual neural networks in the brain than their second-generation counterparts, artificial neural networks (ANNs). SNNs have the potential to be more energy efficient than ANNs on event-driven neuromorphic hardware. This can yield drastic maintenance cost reduction for neural network models, as the energy consumption would be much lower in comparison to regular deep learning models hosted in the cloud today. However, such hardware is still not yet widely available. On standard computer architectures consisting mainly of central processing units (CPUs) and graphics processing units (GPUs) ANNs, due to simpler models of neurons and simpler models of connections between neurons, have the upper hand in terms of execution speed. In general, they also win in terms of learning algorithms, as SNNs do not reach the same levels of performance as their second-generation counterparts in typical machine learning benchmark tasks, such as classification. In this paper, we review existing learning algorithms for spiking neural networks, divide them into categories by type, and assess their computational complexity. [ABSTRACT FROM AUTHOR]

The cooperative aerial and device-to-device (D2D) networks employing non-orthogonal multiple access (NOMA) are expected to play an essential role in next-generation wireless networks. Moreover, machine learning (ML) techniques, such as artificial neural networks (ANN), can significantly enhance network performance and efficiency in fifth-generation (5G) wireless networks and beyond. This paper studies an ANN-based unmanned aerial vehicle (UAV) placement scheme to enhance an integrated UAV-D2D NOMA cooperative network.The proposed placement scheme selection (PSS) method for integrating the UAV into the cooperative network combines supervised and unsupervised ML techniques. Specifically, a supervised classification approach is employed utilizing a two-hidden layered ANN with 63 neurons evenly distributed among the layers. The output class of the ANN is utilized to determine the appropriate unsupervised learning method—either k-means or k-medoids—to be employed. This specific ANN layout has been observed to exhibit an accuracy of 94.12%, the highest accuracy among the ANN models evaluated, making it highly recommended for accurate PSS predictions in urban locations. Furthermore, the proposed cooperative scheme allows pairs of users to be simultaneously served through NOMA from the UAV, which acts as an aerial base station. At the same time, the D2D cooperative transmission for each NOMA pair is activated to improve the overall communication quality. Comparisons with conventional orthogonal multiple access (OMA) and alternative unsupervised machine-learning based-UAV-D2D NOMA cooperative networks show that significant sum rate and spectral efficiency gains can be harvested through the proposed method under varying D2D bandwidth allocations. [ABSTRACT FROM AUTHOR]

Classifications based on deep learning have been widely applied in the estimation of the direction of arrival (DOA) of signal. Due to the limited number of classes, the classification of DOA cannot satisfy the required prediction accuracy of signals from random azimuth in real applications. This paper presents a Centroid Optimization of deep neural network classification (CO-DNNC) to improve the estimation accuracy of DOA. CO-DNNC includes signal preprocessing, classification network, and Centroid Optimization. The DNN classification network adopts a convolutional neural network, including convolutional layers and fully connected layers. The Centroid Optimization takes the classified labels as the coordinates and calculates the azimuth of received signal according to the probabilities of the Softmax output. The experimental results show that CO-DNNC is capable of acquiring precise and accurate estimation of DOA, especially in the cases of low SNRs. In addition, CO-DNNC requires lower numbers of classes under the same condition of prediction accuracy and SNR, which reduces the complexity of the DNN network and saves training and processing time. [ABSTRACT FROM AUTHOR]

Processing-in-Memory (PIM) based on Resistive Random Access Memory (RRAM) is an emerging acceleration architecture for artificial neural networks. This paper proposes an RRAM PIM accelerator architecture that does not use Analog-to-Digital Converters (ADCs) and Digital-to-Analog Converters (DACs). Additionally, no additional memory usage is required to avoid the need for a large amount of data transportation in convolution computation. Partial quantization is introduced to reduce the accuracy loss. The proposed architecture can substantially reduce the overall power consumption and accelerate computation. The simulation results show that the image recognition rate for the Convolutional Neural Network (CNN) algorithm can reach 284 frames per second at 50 MHz using this architecture. The accuracy of the partial quantization remains almost unchanged compared to the algorithm without quantization. [ABSTRACT FROM AUTHOR]

Movie scene event extraction is a practical task in media analysis, which aims at extracting structured events from unstructured movie scripts. However, although there have been many studies regarding open domain event extraction, there have only been a few studies focusing on movie scene event extraction. Specifically aimed at instances where different argument roles have the same characteristics in a movie scene, we propose the utilization of the correlation between different argument roles, which is beneficial for both movie scene trigger extraction (trigger identification and classification) and movie scene argument extraction (argument identification and classification) in event extraction. To model the correlation between different argument roles, we propose the superior role concept (SRC), a high-level role concept based upon the ordinary argument role. In this paper, we introduce a new movie scene event extraction model with two main features: (1) an attentive high-level argument role module to capture SRC information and (2) an SRC-based graph attention network (GAT) to fuse the argument role correlation information into semantic embeddings. To evaluate the performance of our model, we constructed a movie scene event extraction dataset named MovieSceneEvent and also conducted experiments on a widely used dataset to compare the results with other models. The experimental results show that our model outperforms competitive models, and the correlation information of argument roles helps to improve the performance of movie scene event extraction. [ABSTRACT FROM AUTHOR]

The paper discusses the problem of detecting emission sources in a low buildings area using unmanned aerial vehicles. The problem was analyzed, and methods of solving it were presented. Various data acquisition scenarios and their impact on the feasibility of the task were analyzed. A method for detecting smoke objects over buildings using stationary video sequences acquired with a drone in hover with the camera in the nadir position is proposed. The method uses differential frame information from stabilized video sequences and the YOLOv7 classifier. A convolutional network classifier was used to detect the roofs of buildings, using a custom training set adapted to the type of data used. Such a solution, although quite effective, is not very practical for the end user, but it enables the automatic generation of a comprehensive training set for classifiers based on deep neural networks. The effectiveness of such a solution was tested for the latest version of the YOLOv7 classifier. The tests proved the effectiveness of the described method, both for single images and video sequences. In addition, the obtained classifier correctly recognizes objects for sequences that do not meet some of the initial assumptions, such as the angle of the camera capturing the image. [ABSTRACT FROM AUTHOR]

Distance estimation is one of the oldest and most challenging tasks in computer vision using only a monocular camera. This can be challenging owing to the presence of occlusions, noise, and variations in the lighting, texture, and shape of objects. Additionally, the motion of the camera and objects in the scene can affect the accuracy of the distance estimation. Various techniques have been proposed to overcome these challenges, including stereo matching, structured light, depth from focus, depth from defocus, depth from motion, and time of flight. The addition of information from a high-resolution 3D view of the surroundings simplifies the distance calculation. This paper describes a novel distance estimation method that operates with converted point cloud data. The proposed method is a reliable map-based bird's eye view (BEV) that calculates the distance to the detected objects. Using the help of the Euler-region proposal network (E-RPN) model, a LiDAR-to-image-based method for metric distance estimation with 3D bounding box projections onto the image was proposed. We demonstrate that despite the general difficulty of the BEV representation in understanding features related to the height coordinate, it is possible to extract all parameters characterizing the bounding boxes of the objects, including their height and elevation. Finally, we applied the triangulation method to calculate the accurate distance to the objects and statistically proved that our methodology is one of the best in terms of accuracy and robustness. [ABSTRACT FROM AUTHOR]

Piezoelectric actuators (PEA) are high-precision devices used in applications requiring micrometric displacements. However, PEAs present non-linearity phenomena that introduce drawbacks at high precision applications. One of these phenomena is hysteresis, which considerably reduces their performance. The introduction of appropriate control strategies may improve the accuracy of the PEAs. This paper presents a high precision control scheme to be used at PEAs based on the model-based predictive control (MPC) scheme. In this work, the model used to feed the MPC controller has been achieved by means of artificial neural networks (ANN). This approach simplifies the obtaining of the model, since the achievement of a precise mathematical model that reproduces the dynamics of the PEA is a complex task. The presented approach has been embedded over the dSPACE control platform and has been tested over a commercial PEA, supplied by Thorlabs, conducting experiments to demonstrate improvements of the MPC. In addition, the results of the MPC controller have been compared with a proportional-integral-derivative (PID) controller. The experimental results show that the MPC control strategy achieves higher accuracy at high precision PEA applications such as tracking periodic reference signals and sudden reference change. [ABSTRACT FROM AUTHOR]

As the pixel resolution of imaging equipment has grown larger, the images' sizes and the number of pixels used to represent objects in images have increased accordingly, exposing an issue when dealing with larger images using the traditional deep learning models and methods, as they typically employ mechanisms such as increasing the models' depth, which, while suitable for applications that have to be spatially invariant, such as image classification, causes issues for applications that relies on the location of the different features within the images such as object localization and change detection. This paper proposes an adaptive convolutional kernels layer (AKL) as an architecture that adjusts dynamically to images' sizes in order to extract comparable spectral information from images of different sizes, improving the features' spatial resolution without sacrificing the local receptive field (LRF) for various image applications, specifically those that are sensitive to objects and features locations, using the definition of Fourier transform and the relation between spectral analysis and convolution kernels. The proposed method is then tested using a Monte Carlo simulation to evaluate its performance in spectral information coverage across images of various sizes, validating its ability to maintain coverage of a ratio of the spectral domain with a variation of around 20% of the desired coverage ratio. Finally, the AKL is validated for various image applications compared to other architectures such as Inception and VGG, demonstrating its capability to match Inception v4 in image classification applications, and outperforms it as images grow larger, up to a 30% increase in accuracy in object localization for the same number of parameters. [ABSTRACT FROM AUTHOR]

The segmentation of the left ventricle endocardium (LVendo) and the left ventricle epicardium (LVepi) in echocardiography plays an important role in clinical diagnosis. Recently, deep neural networks have been the most commonly used approach for echocardiography segmentation. However, the performance of a well-trained segmentation network may degrade in unseen domain datasets due to the distribution shift of the data. Adaptation algorithms can improve the generalization of deep neural networks to different domains. In this paper, we present a multi-space adaptation-segmentation-joint framework, named MACS, for cross-domain echocardiography segmentation. It adopts a generative adversarial architecture; the generator fulfills the segmentation task and the multi-space discriminators align the two domains on both the feature space and output space. We evaluated the MACS method on two echocardiography datasets from different medical centers and vendors, the publicly available CAMUS dataset and our self-acquired dataset. The experimental results indicated that the MACS could handle unseen domain datasets well, without requirements for manual annotations, and improve the generalization performance by 2.2% in the Dice metric. [ABSTRACT FROM AUTHOR]

Smart grids are able to forecast customers' consumption patterns, i.e., their energy demand, and consequently electricity can be transmitted after taking into account the expected demand. To face today's demand forecasting challenges, where the data generated by smart grids is huge, modern data-driven techniques need to be used. In this scenario, Deep Learning models are a good alternative to learn patterns from customer data and then forecast demand for different forecasting horizons. Among the commonly used Artificial Neural Networks, Long Short-Term Memory networks—based on Recurrent Neural Networks—are playing a prominent role. This paper provides an insight into the importance of the demand forecasting issue, and other related factors, in the context of smart grids, and collects some experiences of the use of Deep Learning techniques, for demand forecasting purposes. To have an efficient power system, a balance between supply and demand is necessary. Therefore, industry stakeholders and researchers should make a special effort in load forecasting, especially in the short term, which is critical for demand response. [ABSTRACT FROM AUTHOR]

In this paper, the useability of feedforward and recurrent neural networks for fusion of data from impulse-radar sensors and depth sensors, in the context of healthcare-oriented monitoring of elderly persons, is investigated. Two methods of data fusion are considered, viz., one based on a multilayer perceptron and one based on a nonlinear autoregressive network with exogenous inputs. These two methods are compared with a reference method with respect to their capacity for decreasing the uncertainty of estimation of a monitored person's position and uncertainty of estimation of several parameters enabling medical personnel to make useful inferences on the health condition of that person, viz., the number of turns made during walking, the travelled distance, and the mean walking speed. Both artificial neural networks were trained on the synthetic data. The numerical experiments show the superiority of the method based on a nonlinear autoregressive network with exogenous inputs. This may be explained by the fact that for this type of network, the prediction of the person's position at each time instant is based on the position of that person at the previous time instants. [ABSTRACT FROM AUTHOR]

This paper presents recent advances in low-latency, single-channel, deep neural network-based speech enhancement systems. The sources of latency and their acceptable values in different applications are described. This is followed by an analysis of the constraints imposed on neural network architectures. Specifically, the causal units used in deep neural networks are presented and discussed in the context of their properties, such as the number of parameters, the receptive field, and computational complexity. This is followed by a discussion of techniques used to reduce the computational complexity and memory requirements of the neural networks used in this task. Finally, the techniques used by the winners of the latest speech enhancement challenges (DNS, Clarity) are shown and compared. [ABSTRACT FROM AUTHOR]

The undeniable computational power of artificial neural networks has granted the scientific community the ability to exploit the available data in ways previously inconceivable. However, deep neural networks require an overwhelming quantity of data in order to interpret the underlying connections between them, and therefore, be able to complete the specific task that they have been assigned to. Feeding a deep neural network with vast amounts of data usually ensures efficiency, but may, however, harm the network's ability to generalize. To tackle this, numerous regularization techniques have been proposed, with dropout being one of the most dominant. This paper proposes a selective gradient dropout method, which, instead of relying on dropping random weights, learns to freeze the training process of specific connections, thereby increasing the overall network's sparsity in an adaptive manner, by driving it to utilize more salient weights. The experimental results show that the produced sparse network outperforms the baseline on numerous image classification datasets, and additionally, the yielded results occurred after significantly less training epochs. [ABSTRACT FROM AUTHOR]

Artifacts are divergent strip artifacts or dark stripe artifacts in Industrial Computed Tomography (ICT) images due to large differences in density among the components of scanned objects, which can significantly distort the actual structure of scanned objects in ICT images. The presence of artifacts can seriously affect the practical application effectiveness of ICT in defect detection and dimensional measurement. In this paper, a series of convolution neural network models are designed and implemented based on preparing the ICT image artifact removal datasets. Our findings indicate that the RF (receptive field) and the spatial resolution of network can significantly impact the effectiveness of artifact removal. Therefore, we propose a dilated residual network for turbine blade ICT image artifact removal (DRAR), which enhances the RF of the network while maintaining spatial resolution with only a slight increase in computational load. Extensive experiments demonstrate that the DRAR achieves exceptional performance in artifact removal. [ABSTRACT FROM AUTHOR]

Reconfigurable Intelligent Surfaces (RISs) not only enable software-defined radio in modern wireless communication networks but also have the potential to be utilized for localization. Most previous works used channel matrices to calculate locations, requiring extensive field measurements, which leads to rapidly growing complexity. Although a few studies have designed fingerprint-based systems, they are only feasible under an unrealistic assumption that the RIS will be deployed only for localization purposes. Additionally, all these methods utilize RIS codewords for location inference, inducing considerable communication burdens. In this paper, we propose a new localization technique for RIS-enhanced environments that does not require RIS codewords for online location inference. Our proposed approach extracts codeword-independent representations of fingerprints using a domain adversarial neural network. We evaluated our solution using the DeepMIMO dataset. Due to the lack of results from other studies, for fair comparisons, we define oracle and baseline cases, which are the theoretical upper and lower bounds of our system, respectively. In all experiments, our proposed solution performed much more similarly to the oracle cases than the baseline cases, demonstrating the effectiveness and robustness of our method. [ABSTRACT FROM AUTHOR]

The accurate identification of the human emotional status is crucial for an efficient human–robot interaction (HRI). As such, we have witnessed extensive research efforts made in developing robust and accurate brain–computer interfacing models based on diverse biosignals. In particular, previous research has shown that an Electroencephalogram (EEG) can provide deep insight into the state of emotion. Recently, various handcrafted and deep neural network (DNN) models were proposed by researchers for extracting emotion-relevant features, which offer limited robustness to noise that leads to reduced precision and increased computational complexity. The DNN models developed to date were shown to be efficient in extracting robust features relevant to emotion classification; however, their massive feature dimensionality problem leads to a high computational load. In this paper, we propose a bag-of-hybrid-deep-features (BoHDF) extraction model for classifying EEG signals into their respective emotion class. The invariance and robustness of the BoHDF is further enhanced by transforming EEG signals into 2D spectrograms before the feature extraction stage. Such a time-frequency representation fits well with the time-varying behavior of EEG patterns. Here, we propose to combine the deep features from the GoogLeNet fully connected layer (one of the simplest DNN models) together with the OMTLBP_SMC texture-based features, which we recently developed, followed by a K-nearest neighbor (KNN) clustering algorithm. The proposed model, when evaluated on the DEAP and SEED databases, achieves a 93.83 and 96.95% recognition accuracy, respectively. The experimental results using the proposed BoHDF-based algorithm show an improved performance in comparison to previously reported works with similar setups. [ABSTRACT FROM AUTHOR]

Currently, the use of Unmanned Aerial Vehicles (UAVs) in natural and complex environments has been increasing, because they are appropriate and affordable solutions to support different tasks such as rescue, forestry, and agriculture by collecting and analyzing high-resolution monocular images. Autonomous navigation at low altitudes is an important area of research, as it would allow monitoring parts of the crop that are occluded by their foliage or by other plants. This task is difficult due to the large number of obstacles that might be encountered in the drone's path. The generation of high-quality depth maps is an alternative for providing real-time obstacle detection and collision avoidance for autonomous UAVs. In this paper, we present a comparative analysis of four supervised learning deep neural networks and a combination of two for monocular depth map estimation considering images captured at low altitudes in simulated natural environments. Our results show that the Boosting Monocular network is the best performing in terms of depth map accuracy because of its capability to process the same image at different scales to avoid loss of fine details. [ABSTRACT FROM AUTHOR]

In this paper, we propose a novel technique for the inspection of high-density polyethylene (HDPE) pipes using ultrasonic sensors, signal processing, and deep neural networks (DNNs). Specifically, we propose a technique that detects whether there is a diversion on a pipe or not. The proposed model transmits ultrasound signals through a pipe using a custom-designed array of piezoelectric transmitters and receivers. We propose to use the Zadoff–Chu sequence to modulate the input signals, then utilize its correlation properties to estimate the pipe channel response. The processed signal is then fed to a DNN that extracts the features and decides whether there is a diversion or not. The proposed technique demonstrates an average classification accuracy of 90.3 % (when one sensor is used) and 99.6 % (when two sensors are used) on 3 4 inch pipes. The technique can be readily generalized for pipes of different diameters and materials. [ABSTRACT FROM AUTHOR]

As the elderly population grows, there is a need for caregivers, which may become unsustainable for society. In this situation, the demand for automated help increases. One of the solutions is service robotics, in which robots have automation and show significant promise in working with people. In particular, household settings and aged people's homes will need these robots to perform daily activities. Clothing manipulation is a daily activity and represents a challenging area for a robot. The detection and classification are key points for the manipulation of clothes. For this reason, in this paper, we proposed to study fashion image classification with four different neural network models to improve apparel image classification accuracy on the Fashion-MNIST dataset. The network models are tested with the highest accuracy with a Fashion-Product dataset and a customized dataset. The results show that one of our models, the Multiple Convolutional Neural Network including 15 convolutional layers (MCNN15), boosted the state of art accuracy, and it obtained a classification accuracy of 94.04% on the Fashion-MNIST dataset with respect to the literature. Moreover, MCNN15, with the Fashion-Product dataset and the household dataset, obtained 60% and 40% accuracy, respectively. [ABSTRACT FROM AUTHOR]

Vehicle dynamics control systems have a fundamental role in smart and autonomous mobility, where one of the most crucial aspects is the vehicle body velocity estimation. In this paper, the problem of a correct evaluation of the vehicle longitudinal velocity for dynamic control applications is approached using a neural networks technique employing a set of measured samples referring to signals usually available on-board, such as longitudinal and lateral acceleration, steering angle, yaw rate and linear wheel speed. Experiments were run on four professional driving circuits with very different characteristics, and the vehicle longitudinal velocity was estimated with different neural network training policies and validated through comparison with the measurements of the one acquired at the vehicle's center of gravity, provided by an optical Correvit sensor, which serves as the reference (and, therefore, exact) velocity values. The results obtained with the proposed methodology are in good agreement with the reference values in almost all tested conditions, covering both the linear and the nonlinear behavior of the car, proving that artificial neural networks can be efficiently employed onboard, thereby enriching the standard set of control and safety-related electronics. [ABSTRACT FROM AUTHOR]

This paper presents the first implementation of a spiking neural network (SNN) for the extraction of cepstral coefficients in structural health monitoring (SHM) applications and demonstrates the possibilities of neuromorphic computing in this field. In this regard, we show that spiking neural networks can be effectively used to extract cepstral coefficients as features of vibration signals of structures in their operational conditions. We demonstrate that the neural cepstral coefficients extracted by the network can be successfully used for anomaly detection. To address the power efficiency of sensor nodes, related to both processing and transmission, affecting the applicability of the proposed approach, we implement the algorithm on specialised neuromorphic hardware (Intel ® Loihi architecture) and benchmark the results using numerical and experimental data of degradation in the form of stiffness change of a single degree of freedom system excited by Gaussian white noise. The work is expected to open a new direction of SHM applications towards non-Von Neumann computing through a neuromorphic approach. [ABSTRACT FROM AUTHOR]

Plant point cloud segmentation is an important step in 3D plant phenotype research. Because the stems, leaves, flowers, and other organs of plants are often intertwined and small in size, this makes plant point cloud segmentation more challenging than other segmentation tasks. In this paper, we propose MASPC_Transform, a novel plant point cloud segmentation network base on multi-head attention separation and position code. The proposed MASPC_Transform establishes connections for similar point clouds scattered in different areas of the point cloud space through multiple attention heads. In order to avoid the aggregation of multiple attention heads, we propose a multi-head attention separation loss based on spatial similarity, so that the attention positions of different attention heads can be dispersed as much as possible. In order to reduce the impact of point cloud disorder and irregularity on feature extraction, we propose a new point cloud position coding method, and use the position coding network based on this method in the local and global feature extraction modules of MASPC_Transform. We evaluate our MASPC_Transform on the ROSE_X dataset. Compared with the state-of-the-art approaches, the proposed MASPC_Transform achieved better segmentation results. [ABSTRACT FROM AUTHOR]

Deep neural networks have been successfully applied to generate predictive patterns from medical and diagnostic data. This paper presents an approach for assessing persons with Alzheimer's disease (AD) mild cognitive impairment (MCI), compared with normal control (NC) persons, using the zoom-in neural network (ZNN) deep-learning algorithm. ZNN stacks a set of zoom-in learning units (ZLUs) in a feedforward hierarchy without backpropagation. The resting-state fMRI (rs-fMRI) dataset for AD assessments was obtained from the Alzheimer's Disease Neuroimaging Initiative (ADNI). The Automated Anatomical Labeling (AAL-90) atlas, which provides 90 neuroanatomical functional regions, was used to assess and detect the implicated regions in the course of AD. The features of the ZNN are extracted from the 140-time series rs-fMRI voxel values in a region of the brain. ZNN yields the three classification accuracies of AD versus MCI and NC, NC versus AD and MCI, and MCI versus AD and NC of 97.7%, 84.8%, and 72.7%, respectively, with the seven discriminative regions of interest (ROIs) in the AAL-90. [ABSTRACT FROM AUTHOR]

Hyperspectral remote sensing images (HRSI) have the characteristics of foreign objects with the same spectrum. As it is difficult to label samples manually, the hyperspectral remote sensing images are understood to be typical "small sample" datasets. Deep neural networks can effectively extract the deep features from the HRSI, but the classification accuracy mainly depends on the training label samples. Therefore, the stacked convolutional autoencoder network and transfer learning strategy are employed in order to design a new stacked convolutional autoencoder network model transfer (SCAE-MT) for the purposes of classifying the HRSI in this paper. In the proposed classification method, the stacked convolutional au-to-encoding network is employed in order to effectively extract the deep features from the HRSI. Then, the transfer learning strategy is applied to design a stacked convolutional autoencoder network model transfer under the small and limited training samples. The SCAE-MT model is used to propose a new HRSI classification method in order to solve the small samples of the HRSI. In this study, in order to prove the effectiveness of the proposed classification method, two HRSI datasets were chosen. In order to verify the effectiveness of the methods, the overall classification accuracy (OA) of the convolutional self-coding network classification method (CAE), the stack convolutional self-coding network classification method (SCAE), and the SCAE-MT method under 5%, 10%, and 15% training sets are calculated. When compared with the CAE and SCAE models in 5%, 10%, and 15% training datasets, the overall accuracy (OA) of the SCAE-MT method was improved by 2.71%, 3.33%, and 3.07% (on average), respectively. The SCAE-MT method is, thus, clearly superior to the other methods and also shows a good classification performance. [ABSTRACT FROM AUTHOR]

This paper focuses on the classification of seven locomotion modes (sitting, standing, level ground walking, ramp ascent and descent, stair ascent and descent), the transitions among these modes, and the gait phases within each mode, by only using data in the frequency domain from one or two inertial measurement units. Different deep neural network configurations are investigated and compared by combining convolutional and recurrent layers. The results show that a system composed of a convolutional neural network followed by a long short-term memory network is able to classify with a mean F 1 -score of 0.89 and 0.91 for ten healthy subjects, and of 0.92 and 0.95 for one osseointegrated transfemoral amputee subject (excluding the gait phases because they are not labeled in the data-set), using one and two inertial measurement units, respectively, with a 5-fold cross-validation. The promising results obtained in this study pave the way for using deep learning for the control of transfemoral prostheses with a minimum number of inertial measurement units. [ABSTRACT FROM AUTHOR]

Using sensors embedded in smartphones to study earthquake early warning (EEW) technology can effectively reduce the high construction and maintenance costs of traditional EEW systems. However, due to the impact of human activities, it is very difficult to accurately detect seismic events recorded on mobile phones. In this paper, to improve the detection accuracy of earthquakes on mobile phones, we investigated the suitability of different types of neural network models in seismic event detection. Firstly, we collected three-component acceleration records corresponding to human activities in various scenarios such as walking, running, and cycling through our self-developed mobile application. Combined with traditional strong-motion seismic event records fusing typical mobile phone accelerometer self-noise, all records were used for establishing the training and testing dataset. Finally, two types of neural network models, fully connected and convolutional neural networks, were trained, validated, and tested. The results showed that the accuracy rates of the neural network models were all over 98%, and the precision rate for seismic events and the recall rate for non-earthquake events could both reach 99%, indicating that the introduction of neural networks into the earthquake recognition on smartphones can significantly enhance the accuracy of seismic event recognition. Therefore, we can exceedingly reduce the amount of data transmitted to the processing server, further lowering the load on the server processor and effectively increasing the lead time at each target site for an EEW system. [ABSTRACT FROM AUTHOR]

Although gait recognition has been greatly improved by efforts from many researchers in recent years, its performance is still unsatisfactory due to the lack of gait information under the real scenariowhere only one or two images may be used for recognition. In this paper, a new gait recognition framework is brought about which can combine the long-short-term attention modules on silhouette images over the whole sequence and the real human physiological information calculated by a monocular image. The contributions of this work include the following: (1) Fusing the global long-term attention (GLTA) and local short-term attention (LSTA) over the whole query sequence to improve the gait recognition accuracy, where both the short-term gait feature (from two or three frames) and long-term feature (from the whole sequence) are extracted; (2) presenting a method to calculate the real personal static and dynamic physiological features through a single monocular image; (3) by efficiently applying the human physiological information, a new physiological feature extraction (PFE) network is proposed to concatenate the physiological information with silhouette for gait recognition. Through the experiments between the CASIA-B and Multi-state Gait datasets, the effectiveness and efficiency of the proposed method are proven. Under three different walking conditions of the CASIA-B dataset, the mean accuracy of rank-1 in our method is up to 89.6%, and in the Multi-state Gait dataset, wearing different clothes, the mean accuracy of rank-1 in our method is 2.4% higher than the other works. [ABSTRACT FROM AUTHOR]

Currently, glaucoma has become an important cause of blindness. At present, although glaucoma cannot be cured, early treatment can prevent it from getting worse. A reliable way to detect glaucoma is to segment the optic disc and cup and then measure the cup-to-disc ratio (CDR). Many deep neural network models have been developed to autonomously segment the optic disc and the optic cup to help in diagnosis. However, their performance degrades when subjected to domain shift. While many domain-adaptation methods have been exploited to address this problem, they are apt to produce malformed segmentation results. In this study, it is suggested that the segmentation network be adjusted using a constrained formulation that embeds prior knowledge about the shape of the segmentation areas that is domain-invariant. Based on IOSUDA (i.e., Input and Output Space Unsupervised Domain Adaptation), a novel unsupervised joint optic cup-to-disc segmentation framework with shape constraints is proposed, called SCUDA (short for Shape-Constrained Unsupervised Domain Adaptation). A shape constrained loss function is novelly proposed in this paper which utilizes domain-invariant prior knowledge concerning the segmentation region of the joint optic cup–optical disc of fundus images to constrain the segmentation result during network training. In addition, a convolutional triple attention module is designed to improve the segmentation network, which captures cross-dimensional interactions and provides a rich feature representation to improve the segmentation accuracy. Experiments on the RIM-ONE_r3 and Drishti-GS datasets demonstrate that the algorithm outperforms existing approaches for segmenting optic discs and cups. [ABSTRACT FROM AUTHOR]

Recent work has shown that deep neural networks are vulnerable to backdoor attacks. In comparison with the success of backdoor-attack methods, existing backdoor-defense methods face a lack of theoretical foundations and interpretable solutions. Most defense methods are based on experience with the characteristics of previous attacks, but fail to defend against new attacks. In this paper, we propose IBD, an interpretable backdoor-detection method via multivariate interactions. Using information theory techniques, IBD reveals how the backdoor works from the perspective of multivariate interactions of features. Based on the interpretable theorem, IBD enables defenders to detect backdoor models and poisoned examples without introducing additional information about the specific attack method. Experiments on widely used datasets and models show that IBD achieves a 78 % increase in average in detection accuracy and an order-of-magnitude reduction in time cost compared with existing backdoor-detection methods. [ABSTRACT FROM AUTHOR]

The early detection of pathogen infections in plants has become an important aspect of integrated disease management. Although previous research demonstrated the idea of applying digital technologies to monitor and predict plant health status, there is no effective system for detecting pathogen infection before symptomatology appears. This paper presents the use of a low-cost and portable electronic nose coupled with machine learning (ML) models for early disease detection. Several artificial neural network models were developed to predict plant physiological data and classify processing tomato plants and soil samples according to different levels of pathogen inoculum by using e-nose outputs as inputs, plant physiological data, and the level of infection as targets. Results showed that the pattern recognition models based on different infection levels had an overall accuracy of 94.4–96.8% for tomato plants and between 94.81% and 96.22% for soil samples. For the prediction of plant physiological parameters (photosynthesis, stomatal conductance, and transpiration) using regression models or tomato plants, the overall correlation coefficient was 0.97–0.99, with very significant slope values in the range 0.97–1. The performance of all models shows no signs of under or overfitting. It is hence proven accurate and valid to use the electronic nose coupled with ML modeling for effective early disease detection of processing tomatoes and could also be further implemented to monitor other abiotic and biotic stressors. [ABSTRACT FROM AUTHOR]

Contributions in the Field of Image Techniques Five papers in this Special Issue deal with image processing, including face images, microscopic images, and infrared images. Nowadays, with numerous sensors placed everywhere around us, we can obtain signals collected from a variety of environment-based sensors, including the ones placed on the ground, cased in the air or water, etc. M. Geremek and K. Szklanny in [[9]] investigated deep learning based detection of genetic diseases from face images, for 15 genetic disorders associated with facial dysmorphism. Image and video data analyzed in the presented papers include microscope images and infrared images, as well as face images. [Extracted from the article]

Deep neural networks have been deployed in various hardware accelerators, such as graph process units (GPUs), field-program gate arrays (FPGAs), and application specific integrated circuit (ASIC) chips. Normally, a huge amount of computation is required in the inference process, creating significant logic resource overheads. In addition, frequent data accessions between off-chip memory and hardware accelerators create bottlenecks, leading to decline in hardware efficiency. Many solutions have been proposed to reduce hardware overhead and data movements. For example, specific lookup-table (LUT)-based hardware architecture can be used to mitigate computing operation demands. However, typical LUT-based accelerators are affected by computational precision limitation and poor scalability issues. In this paper, we propose a search-based computing scheme based on an LUT solution, which improves computation efficiency by replacing traditional multiplication with a search operation. In addition, the proposed scheme supports different precision multiple-bit widths to meet the needs of different DNN-based applications. We design a reconfigurable computing strategy, which can efficiently adapt to the convolution of different kernel sizes to improve hardware scalability. We implement a search-based architecture, namely SCA, which adopts an on-chip storage mechanism, thus greatly reducing interactions with off-chip memory and alleviating bandwidth pressure. Based on experimental evaluation, the proposed SCA architecture can achieve 92%, 96% and 98% computational utilization for computational precision of 4 bit, 8 bit and 16 bit, respectively. Compared with state-of-the-art LUT-based architecture, the efficiency can be improved four-fold. [ABSTRACT FROM AUTHOR]