Search

This paper aims to craft adversarial queries for image retrieval, which uses image features for similarity measurement. Many commonly used methods are developed in the context of image classification. However, these methods, which attack prediction probabilities, only exert an indirect influence on the image features and are thus found less effective when being applied to the retrieval problem. In designing an attack method specifically for image retrieval, we introduce opposite-direction feature attack (ODFA), a white-box attack approach that directly attacks query image features to generate adversarial queries. As the name implies, the main idea underpinning ODFA is to impel the original image feature to the opposite direction, similar to a U-turn. This simple idea is experimentally evaluated on five retrieval datasets. We show that the adversarial queries generated by ODFA cause true matches no longer to be seen at the top ranks, and the attack success rate is consistently higher than classifier attack methods. In addition, our method of creating adversarial queries can be extended for multi-scale query inputs and is generalizable to other retrieval models without foreknowing their weights, i.e., the black-box setting. [ABSTRACT FROM AUTHOR]

In this paper, a method based on transfer learning is proposed to recover the three-dimensional shape of cultural relics faces from a single old photo. It can simultaneously reconstruct the three-dimensional facial structure and align the texture of the cultural relics with fewer training samples. The UV position map is used to represent the three-dimensional shape in space and act as the output of the network. A convolutional neural network is used to reconstruct the UV position map from a single 2D image. In the training process, the human face data is used for pre-training, and then a small amount of artifact data is used for fine-tuning. A deep learning model with strong generalization ability is trained with fewer artifact data, and a three-dimensional model of the cultural relic face can be reconstructed from a single old photograph. The methods can train more complex deep networks without a large amount of cultural relic data, and no over-fitting phenomenon occurs, which effectively solves the problem of fewer cultural relic samples. The method is verified by restoring a Chinese Terracotta Warrior with facial cracks and a Buddha with a broken nose. Other applications can be used in the fields such as texture recovery, facial feature extraction, and three-dimensional model estimation of the damaged cultural relics or sculptures in the photos. [ABSTRACT FROM AUTHOR]

Fault diagnosis of industrial processes plays an important role in avoiding heavy losses and ensuring production safety. Complex industrial processes often have many working conditions, and the actual industrial process often concentrates on certain working conditions. As a result, the running time of some working conditions is shorter, so the data of these conditions are difficult to obtain. However, almost all fault diagnosis methods based on Deep Learning (DL) requires a large amount of data. Therefore, it is a big challenge to realize the fault diagnosis of few-shot industrial processes. In order to solve these problems, this paper proposes a Deep Feature Transfer Fusion (DFTF) framework based on hybrid domain transfer learning. The purpose is to take few-shot working conditions as the target domain and carry out fault diagnosis for them. As the features of industrial process images are more complex, this paper introduces Pyramid Hybrid Vision Transformer (PHVT) model, which have stronger feature extraction capabilities and spatial perception, as feature extraction module. In order to improve the transferability of the model, this paper introduces the In-Cross Domain Hybrid Transfer Learning (ICTL) method. By fusing the general object features from ResNet50 which pre-trained under public dataset of common object and the features extracted from PHVT which pre-trained under industrial dataset of multiple working conditions, the adaptability of the model to different scenes is enhanced. The experimental results based on Pronto dataset of Process System Engineering lab of Cranfield University show that the proposed transfer learning method has excellent performance. [ABSTRACT FROM AUTHOR]

Furnace temperature and oxygen content are important parameters reflecting the combustion inside a circulating fluidized bed (CFB) boiler. Accurately predicting boiler output is a complex task due to the high noise and nonsmoothness of actual boiler input and output data. In this paper, a new hybrid convolutional neural network (CNN), bidirectional long short-term memory (biLSTM) network, and squeezing and excitation (SE) network prediction model is proposed to significantly improve the prediction accuracy of oxygen content and furnace temperature by combining the advantages of multiple deep learning networks. This network can extract spatiotemporal characteristics of input parameters such as coal feed to effectively predict boiler furnace temperature and oxygen content. CNNs can extract complex features such as dynamic and static nonlinearities between multiple variables affecting the furnace temperature and oxygen content, as well as high noise. The biLSTM network layer can efficiently handle the temporal information of irregular trends in modeling time series components; SE can extract the important information between channels through the feature relationships between channels for better overfeature extraction. The CNN-biLSTM-SE model can effectively solve the problem of nonlinear mapping complexity between inputs and outputs. Experiments show that the proposed CNN-biLSTM-SE model outperforms existing methods. The experimental results showed that the average MAPE errors for oxygen content prediction were CNN-biLSTM-SE (0.038), CNN-biLSTM with attention mechanism (AM) (0.043), CNN-biLSTM (0.051), CNN-LSTM (0.051), biLSTM (0.051), RNN (0.051), LSTM(0.0052), and CNN(0.0054). Extensive experiments in CFB boilers with oxygen content and furnace temperature show that the proposed CNN-biLSTM-SE model achieves better results in terms of goodness-of-fit, generalization ability and accuracy. [ABSTRACT FROM AUTHOR]

To achieve scientific and intelligent hierarchical intervention of violent criminals, this paper proposes a deep feature fusion model called the Convolutional Long Short Term Gated Attention Network (CLGA-Net), for violent crime temperament classification. First, the CNN is reconstructed to improve local feature extraction. We combine eigenvectors by filtering word vectors with multiple kernel sizes and convoluting separately to obtain feature maps with different granularities. Second, a multi-model fusion algorithm with attention mechanism is proposed by reconstructing CNN. After concatenate layer, Bi-LSTM and Bi-GRU are combined with Multi-Head Attention, and parallelly combined LSTM and GRU with Self-Attention to capture various structural patterns effectively in text sequences. Then, global average pooling layer is used to solve the overfitting of the traditional fully connected layer. Finally, after the softmax classifier, the arithmetic average algorithm is used to obtain the final classification result. To verify the performance of the proposed model, CLGA-Net was compared with other baseline models. The results show that CLGA-Net achieves the optimal effect under all evaluation indices with an accuracy of 99.25%. The AUC values under the macro average and micro average are 99.03% and 99.81%, respectively. By analyzing crime facts, CLGA-Net shows good violent crime temperament classification ability. Regulatory departments can develop a personalized correctional education program according to the type of attribution obtained. Therefore, the CLGA-Net model proposed in this paper can provide an accurate and scalable computerized hierarchical intervention service for offenders. [ABSTRACT FROM AUTHOR]

Collecting and analyzing data from all devices to improve the efficiency of business processes is an important task of Industrial Internet of Things (IIoT). In the age of data explosion, extensive text data generated by the IIoT have given birth to a variety of text representation methods. The task of text representation is to convert the natural language to a form that computer can understand with retaining the original semantics. However, these methods are difficult to effectively extract the semantic features among words and distinguish polysemy in natural language. Combining the advantages of convolutional neural network (CNN) and variational autoencoder (VAE), this paper proposes an intelligent CNN-VAE text representation algorithm as an advanced learning method for social big data within next-generation IIoT, which help users identify the information collected by sensors and perform further processing. This method employs the convolution layer to capture the local features of the context and uses the variational technique to reconstruct feature space to make it conform to the normal distribution. In addition, the improved word2vec model based on topical word embedding (TWE) is utilized to add topical information to word vectors to distinguish polysemy. This paper takes the social big data as an example to illustrate the way of the proposed algorithm applied in the next-generation IIoT and utilizes Cnews dataset to verify the performance of proposed method with four evaluating metrics (i.e., recall, accuracy, precision, and F1-score). Experimental results indicate that the proposed method outperforms word2vec-avg and CNN-AE in K-nearest neighbor (KNN), random forest (RF), and support vector machine (SVM) classifiers and distinguishes polysemy effectively. [ABSTRACT FROM AUTHOR]

Sleep is essential for human physical and mental health. Sleep disorders are a significant threat to human health, and a large number of people in the world suffer from sleep disorders. Effective detection of sleep disorders is essential for the treatment of sleep disorders. Questionnaires and scale assessments are traditional methods of sleep disorder detection, which are subjective, time-consuming and prone to misdiagnosis. To detect sleep disorders quickly and accurately, a Multimodal Attention-Fusion Convolutional Neural Network is proposed in this paper. The network uses electroencephalography, electrooculography, electrocardiography, and electromyography signals to automatically identify healthy and five sleep disorders, namely insomnia, narcolepsy, periodic leg movement, rapid eye movement behaviour disorder and nocturnal frontal lobe epilepsy. First, multiple convolutional neural network branches are used to extract time-invariant features of multimodal signals. Then, a multi-scale attention module based on dilated convolutional networks and a squeeze and excite block is proposed for further extracting features with different scales and fusing feature information. Finally, a prediction module consisting of fully connected layers is used to detect sleep disorders. The accuracy, F1 score, and Kappa coefficient obtained on the Cyclic Alternating Pattern sleep dataset are 99.56%, 99.49% and 0.9942, respectively. Compared to the existing state-of-the-art studies, the method proposed in this paper has higher performance in sleep disorder detection. [ABSTRACT FROM AUTHOR]

In recent years, the field of bionics has attracted the attention of numerous scholars. Some models combined with biological vision have achieved excellent performance in computer vision and image processing tasks. In this paper, we propose a new bio-inspired lightweight contour detection network (BLCDNet) by combining parallel processing mechanisms of bio-visual information with convolutional neural networks. The backbone network of BLCDNet simulates the parallel pathways of ganglion cell–lateral geniculate nucleus and primary visual cortex (V1) area, realizing parallel processing and step-by-step extraction of input information, effectively extracting local features and detailed features in images, and thus improving the overall performance of the model. In addition, we design a depth feature extraction module combining depth separable convolution and residual connection in the decoding network to integrate the output of the backbone network, which further improves the performance of the model. We conducted a large number of experiments on BSDS500 and NYUD datasets, and the experimental results show that the BLCDNet proposed in this paper achieves the best performance compared with traditional methods and previous biologically inspired contour detection methods. In addition, BLCDNet still outperforms some VGG-based contour detection methods without pre-training and with fewer parameters, and it is competitive among all of them. The research in this paper also provides a new idea for the combination of biological vision and convolutional neural networks. [ABSTRACT FROM AUTHOR]

With the recent advancements in high-performance computing, convolutional neural networks (CNNs) have achieved remarkable success in various vision tasks. However, along with improvements in model accuracy, the size and computational complexity of the models have significantly increased with the increasing number of parameters. Although graphics processing unit (GPU) platforms equipped with high-performance memory and specialized in parallel processing are commonly used for CNN processing, the significant power consumption presents challenges in their utilization on edge devices. To address these issues, research is underway to design CNN models using field-programmable gate arrays (FPGAs) as accelerators. FPGAs provide a high level of flexibility, allowing efficient optimization of convolution operations, which account for a significant portion of the CNN computations. Additionally, FPGAs are known for their low power consumption compared to GPUs, making them a promising energy-efficient platform. In this paper, we review and summarize various approaches and techniques related to the design of FPGA-based CNN accelerators. Specifically, to comprehensively study CNN accelerators, we investigate the advantages and disadvantages of various methods for optimizing CNN accelerators and previously designed efficient accelerator architectures. We expect this paper to serve as an important guideline for future hardware research in artificial intelligence. [ABSTRACT FROM AUTHOR]

Time series analysis is widely applied in action recognition, anomaly detection, and weather forecasting. Time series forecasting remains a key challenge due to the complexity of temporal patterns, overlapping changes within sequences, and the need for advanced predictive models to forecast longer sequences in many scenarios. In this study, a model named the decomposed dimension time-domain convolutional neural network (DDTCN) is proposed. This model is specifically designed to address the challenges associated with long time series data. This paper presents a dimension temporal convolutional network (DTCN) module, which has a strong ability to capture variable correlations, and an adaptive strategy is introduced. Specifically, the model proposed in this paper first decomposes time series trends and is combined with the DTCN to extract variable correlations, thus achieving accurate predictions for complex time series and providing a powerful solution for long-term series forecasting. Experiments are conducted on multiple long-term series datasets covering five practical applications: energy, transportation, economics, weather, and health care. The proposed model is extensively evaluated and compared with traditional time series prediction methods and several benchmarks. The experimental results demonstrate that the proposed model outperforms state-of-the-art methods in most tasks involving multiple long-term series forecasting. Additionally, a pig price dataset is generated to predict agricultural economic trends, where compared to that of state-of-the-art methods, the DDTCN achieves a reduction in prediction error of 25.36%. Hence, this model holds promising prospects for wide-ranging applications. [ABSTRACT FROM AUTHOR]

Accurate and stable estimation of the state of health (SOH), which is one of the critical indicators to characterize the ability of lithium-ion (Li-ion) batteries to store and release energy, is critical in the stable driving of electric vehicles. In this paper, a novel SOH estimation method based on the aging factors of battery, which combines convolutional neural network (CNN), wavelet neural network (WNN), and wavelet long short-term memory (WLSTM) named CNN–WNN–WLSTM, is designed. The proposed CNN–WNN–WLSTM estimation scheme inherits both the fast convergence and robust stability of the WNN, as well as the ability of long short-term memory neural network (LSTM) to extract the time series features of the data; moreover, using CNN can make the proposed algorithm extract the data features from the original battery data automatically, and the WNN–WLSTM is then adopted to produce the final SOH estimation by exploiting the features from the CNN. To further speed and achieve global optimization, the RMSprop optimizer, instead of the usually used Adagrad optimizer, is chosen as the solver of the CNN–WNN–WLSTM network. Experimental results on data set from the NASA Ames Prognostics Center of Excellence show that the proposed algorithm can be commendably used for Li-ion battery health management by quantitative comparison with other commonly used machine learning methods, such as back-propagation neural network, WNN, LSTM, WLSTM, convolutional neural network–long short-term memory neural network (CNN–LSTM), and Gaussian process regression. [ABSTRACT FROM AUTHOR]

Learning effective feature spaces for KNN (K-Nearest Neighbor) classifiers is critical for their performance. Existing KNN loss functions designed to optimize CNNs in R n feature spaces for specific KNN classifiers greatly boost the performance. However, these loss functions need to compute the pairwise distances within each batch, which requires large computational resource, GPU and memory. This paper aims to exploit lightweight KNN loss functions in order to reduce the computational cost while achieving comparable to or even better performance than existing KNN loss functions. To this end, an anchor loss function is proposed that assigns each category an anchor vector in KNN feature spaces and introduces the distances between training samples and anchor vectors in the NCA (Neighborhood Component Analysis) function. The proposed anchor loss function largely reduces the required computation by existing KNN loss functions. In addition, instead of optimizing CNNs in R n feature spaces, this paper proposed to optimize them in hypersphere feature spaces for faster convergence and better performance. The proposed anchor loss optimized in the hypersphere feature space is called HAL (Hypersphere Anchor Loss). Experiments on various image classification benchmarks show that HAL reduces the computational cost and achieves better performance: on CIFAR-10 and Fashion-MNIST datasets, compared with existing KNN loss functions, HAL improves the accuracy by over 1 % , and the computational cost decreases to less than 10 % . [ABSTRACT FROM AUTHOR]

Distance calculation is a critical link in the research fields of object trajectory prediction, automatic driving obstacle avoidance, and so on. However, the research on distance using deep learning methods has yet to attract wide attention. The accuracy of traditional distance estimation algorithms based on the optical principle and mathematical modeling is low in practical applications, mainly the curve or slope of the road surface. This paper addresses the challenging distance estimation problem by developing an end-to-end structured model to directly predict the distance for objects in a given image. Besides, the traditional mathematical modeling process is replaced by this learning-based method. To facilitate the research on this task, we construct the extended distance datasets by KITTI (Karlsruhe Institute of Technology and Toyota Technological Institute) and NYU(Nathan Silberman, Pushmeet Kohli, Derek Hoiem, Rob Fergus) Depth V2 distance datasets. Experimental results demonstrate that the structured learning model has higher accuracy than the traditional algorithm in different distance ranges and better performance for curves and ramps. Moreover, improving neural network performance will be the direction of improving the model in the future. [ABSTRACT FROM AUTHOR]

Convolutional neural networks (CNNs) have been widely used for object recognition and grasping posture planning in intelligent robotic grasping (IRG). Compared with the traditional usage of CNNs in image recognition, IRGs require high recognition accuracy and computational efficiency. However, the existing methodologies for CNN architecture design often rely on human experience and numerous trial-and-error attempts, which make it a very challenging task to obtain an optimal CNN for IRGs. To tackle this challenge, this paper develops a new differentiable architecture search (DARTS) method considering the floating-point operations (FLOPs) of CNNs, named the DARTS-F method, which converts the discrete CNN architecture search to a gradient-based continuous optimization problem and considers both the prediction accuracy and the computational cost of the CNN during the optimization. To efficiently identify the optimal neural network, this paper adopts a bilevel optimization, which first trains the neural network weights in the inner level and then optimizes the neural network architecture by fine-tuning the operational variables in the outer level. In addition, a new digital twin (DT) of IRG is developed considering the physics of realistic robotic grasping in the DT's virtual space, which could not only improve the IRG accuracy but also avoid the expensive training time. In the experiments, the proposed DARTS-F method could generate an optimized CNN with higher prediction accuracy and lower FLOPs than those obtained by the original DARTS method. The DT framework improves the accuracy of real robotic grasping from 61 to 71%. [ABSTRACT FROM AUTHOR]

When the gearbox fails, the impulsive fault feature is usually submerged in other frequency component. The recognition accuracy of the network model will be seriously affected when directly using the vibration signal as the sample input of Convolutional Neural Network (CNN). To solve this problem, a state recognition method based on vibration signal fusion feature convolutional learning network (FF-CNN) is proposed to conduct the gear fault diagnosis. This method uses the Resonance-based Sparse Signal Decomposition (RSSD) algorithm to separate the periodic and impulsive components that characterize different fault characteristics of the vibration signal. The impulsive feature is amplified by the Teager energy operator. And then the signal periodic and impulsive components are input into CNN to perform the targeted deep learning. Finally, the efficiency of this method is validated using the gear power transmission failure simulation experimental setup. The research in this paper improves the effectiveness of the feature learning, reduces the complexity of deep learning model, and ensures the accuracy of state recognition. Through experimental research, it is found that the recognition accuracy of the recognition model based on the method in this paper reaches more than 95.6%. Highlights: The experiments were carried out in the dynamic engineering laboratory. An FF-CNN method is proposed to conduct the fault diagnosis. T-SNE tool in the manifold learning method was used to visualize the results. FF-CNN can improve the accuracy of fault diagnosis by separating fault features. [ABSTRACT FROM AUTHOR]

To solve the problem that existing traffic signs are not easily detected leading to low detection performance due to their small sizes and external factors such as weather conditions, this paper proposes a traffic sign detection method, MTSDet (Multi-scale Traffic Sign Detection with attention and path aggregation), which focuses on the multi-scale detection problem and effectively improves the detection performance. First, the method efficiently extracts semantic features by introducing the Attention Mechanism Network(AMNet), and then feeds the multi-scale semantic features into Path Aggregation Feature Pyramid Network(PAFPN) for multi-scale feature fusion to obtain multi-scale advanced semantic features. Finally, the multi-scale advanced semantic feature map is deformable interest pooled to effectively enhance the multi-scale object detection modeling capability. In this paper, the above method is validated by two classical datasets, German traffic sign detection dataset and Chinese traffic sign detection dataset, which achieve 92.9% and 94.3% mAP, respectively, and have obvious detection accuracy improvement when compared with other classical advanced algorithms, effectively proving the superiority and generalization of the algorithm in this paper. Code is available at https://github.com/why529913/MTSDet [ABSTRACT FROM AUTHOR]

The coronavirus first appeared in China in 2019, and the World Health Organization (WHO) named it COVID-19. Then WHO announced this illness as a worldwide pandemic in March 2020. The number of cases, infections, and fatalities varied considerably worldwide. Because the main characteristic of COVID-19 is its rapid spread, doctors and specialists generally use PCR tests to detect the COVID-19 virus. As an alternative to PCR, X-ray images can help diagnose illness using artificial intelligence (AI). In medicine, AI is commonly employed. Convolutional neural networks (CNN) and deep learning models make it simple to extract information from images. Several options exist when creating a deep CNN. The possibilities include network depth, layer count, layer type, and parameters. In this paper, a novel Xception-based neural network is discovered using the genetic algorithm (GA). GA finds better alternative networks and parameters during iterations. The best network discovered with GA is tested on a COVID-19 X-ray image dataset. The results are compared with other networks and the results of papers in the literature. The novel network of this paper gives more successful results. The accuracy results are 0.996, 0.989, and 0.924 for two-class, three-class, and four-class datasets, respectively. [ABSTRACT FROM AUTHOR]

Deep fake technology has emerged as a double-edged sword in the digital world. While it holds potential for legitimate uses, it can also be exploited to manipulate video content, causing severe social and security concerns. The research gap lies in the fact that traditional deep fake detection methods, such as visual quality analysis or inconsistency detection, need help to keep up with the rapidly advancing technology used to create deep fakes. That means there's a need for more sophisticated detection techniques. This paper introduces an enhanced approach for detecting deep fake videos using graph neural network (GNN). The proposed method splits the detection process into two phases: a mini-batch graph convolution network stream four-block CNN stream comprising Convolution, Batch Normalization, and Activation function. The final step is a flattening operation, which is essential for connecting the convolutional layers to the dense layer. The fusion of these two phases is performed using three different fusion networks: FuNet-A (additive fusion), FuNet-M (element-wise multiplicative fusion), and FuNet-C (concatenation fusion). The paper further evaluates the proposed model on different datasets, where it achieved an impressive training and validation accuracy of 99.3% after 30 epochs. [ABSTRACT FROM AUTHOR]

With the development of high-resolution camera technology, the shooting scene coverage has reached the square kilometer level, thousands of people can be observed at the same time, and the faces of people from a hundred meters away are clearly recognizable. The images captured by high-resolution cameras are very different from those captured by conventional cameras. In the face of many detection targets in high-resolution images, large differences in target scales due to spatial position, as well as difficulties in extracting features and poor detection results caused by target overlap and concealment phenomena, this paper proposes a multi-target detection method SARNet that combined with spatial attention optimization feature extraction. Use spatial attention to optimize the backbone network, expand the local receptive field, thereby enhance the representation ability, and enhance the feature extraction ability of small targets; the different scale features of the dilated feature pyramid network are subjected to the deformable region of interest pooling operation, which effectively improves the different scales detection accuracy. The experimental results show that the method proposed in this paper can get 51.9% mAP on the PANDA dataset, which is superior to the existing detection algorithms. At the same time, experimental verification of pedestrians and vehicles on the COCO2017 dataset fully proves the feasibility of the method in this paper. [ABSTRACT FROM AUTHOR]

Blockchain technology has unlocked the doors of building decentralized applications, where security plays a vital role. Any transaction ever created in a blockchain is recorded permanently. With the increasing usage of social media modalities today, many non-reputable sources create and publish fake and luring news. Due to the easy availability of the internet and excessive use of social media, fake news can spread like a flash. It causes the need to eliminate fake news posted on social media platforms. The main motive of the paper is to develop an ensemble model that assists in the automated identification of fake news. This paper addresses the identified gaps in detecting multi-modal fake news. We propose Fake News Network (FNNet) as a novel secure blockchain-based deep learning model to detect fake news. It uses blockchain and deep learning to assure data integrity and learn data representations, respectively. The proposed model consists of four layers: the node layer, the deep learning layer, the blockchain layer, and the network layer. This is an ensemble deep-learning model that leverages Bi-LSTM and CNN models, where the Bi-LSTM model captures the sequential data in both directions, and CNN captures the hidden features. We use Pheme, CrisisLexT6, and ISOT datasets to train the model. FNNet achieves an average accuracy of 86.93%, 92.22%, and 98.53% on Pheme, CrisisLexT9, and ISOT datasets, respectively. Our results show that the proposed model is robust and applicable to real-time datasets and social media networks to detect fake news effectively. [ABSTRACT FROM AUTHOR]

Aiming at the complex background of transmission lines at the present stage, which leads to the problem of low accuracy of insulator fault detection for small targets, a deep learning-based insulator fault detection algorithm for transmission lines is proposed. First, aerial images of insulators are collected using UAVs in different scenarios to establish insulator fault datasets. After that, in order to improve the detection efficiency of the target detection algorithm, certain improvements are made on the basis of the YOLOV9 algorithm. The improved algorithm enhances the feature extraction capability of the algorithm for insulator faults at a smaller computational cost by adding the GAM attention mechanism; at the same time, in order to realize the detection efficiency of small targets for insulator faults, the generalized efficient layer aggregation network (GELAN) module is improved and a new SC-GELAN module is proposed; the original loss function is replaced by the effective intersection-over-union (EIOU) loss function to minimize the difference between the aspect ratio of the predicted frame and the real frame, thereby accelerating the convergence speed of the model. Finally, the proposed algorithm is trained and tested with other target detection algorithms on the established insulator fault dataset. The experimental results and analysis show that the algorithm in this paper ensures a certain detection speed, while the algorithmic model has a higher detection accuracy, which is more suitable for UAV fault detection of insulators on transmission lines. [ABSTRACT FROM AUTHOR]

The accurate prediction of a lithium-ion battery's State of Health is of critical importance for efficient battery health management. Existing data-driven estimation methodologies grapple with issues such as high model complexity and a dearth of guidance from prior knowledge, which impose constraints on their efficacy. This work introduces a novel cross-modal distillation network for battery State of Health estimation, structured around a TransformerEncoder as the teacher network and a Convolutional Neural Network as the student network. Initially, the teacher model is pre-trained offline using State of Health degradation data to learn the degradation patterns. The directly measurable feature data (such as voltage, temperature, and current) is subsequently fed into the student network for online training and computation of a hard loss. the student network's output is then directed into the pre-trained the teacher network to compute a soft loss, thereby offering prior knowledge of degradation laws and steering the optimization process of the student network. Rigorous experiments are conducted utilizing various datasets, with the outcomes validating the superior estimation accuracy and degradation rule adherence of the model. Notably, among five different models, this model demonstrates the best performance on almost all datasets, achieving an RMSE of 0.0097 and an MAE of 0.0065 on Cell1 of the Oxford dataset. Moreover, the model also demonstrates robust performance across different usage scenarios, inclusive of multi-battery estimation. Furthermore, this paper also introduces a fine tuning method for State of Health predictions only using the first half of the data. Comparative analysis with other models underscores the competitiveness of the proposed model, showcasing its potential for broader application. [ABSTRACT FROM AUTHOR]

Object detection plays a vital role in remote sensing applications. Although object detection has achieved proud results in natural images, these methods are difficult to be directly applied to remote sensing images. Remote sensing images often have complex backgrounds and small objects, which results in a highly unbalanced distribution of foreground and complex background information. In order to solve the above problems, this paper proposes a multi-head channel and spatial trans-attention (MCSTA) module, which performs remote pixel interaction from the channel and spatial dimensions respectively to complete the attention feature capture function. It is a plug-and-play module that can be easily embedded in any other natural image object detection convolutional neural network, making it quickly applicable to remote sensing images. First, in order to reduce computational complexity and improve feature richness, we use a special linear convolution to obtain three projection features instead of the simple matrix multiplication transformation in Transformer. Second, we obtain trans-attention maps in different dimensions in a manner similar to the self-attention mechanism to capture the interrelationships of features in channels and spaces. In this process, we use a multi-head mechanism to perform parallel operations to improve speed. Furthermore, in order to avoid large-scale matrix operations, we specially designed an attention blocking mode to reduce computer memory usage and increase operation speed. Finally, we embedded the trans-attention module into YOLOv8, added a new detection head and optimized the feature fusion method, thus designing a lightweight small object detection model named TA-YOLO for remote sensing images. It has fewer parameters than the benchmark model YOLOv8, and its mAP on the PASCAL VOC and VisDrone data sets increased by 1.3% and 6.2% respectively. The experimental results prove the powerful function of the trans-attention module and the excellent performance of TA-YOLO. [ABSTRACT FROM AUTHOR]

The integration of convolutional neural network (CNN) and transformer enhances the network's capacity for concurrent modeling of texture details and global structures. However, training challenges with transformer limit their effectiveness to low-resolution images, leading to increased artifacts in slightly larger images. In this paper, we propose a single-stage network utilizing large kernel attention (LKA) to address high-resolution damaged images. LKA enables the capture of both global and local details, akin to transformer and CNN networks, resulting in high-quality inpainting. Our method excels in: (1) reducing parameters, improving inference speed, and enabling direct training on 1024 × 1024 resolution images; (2) utilizing LKA for enhanced extraction of global high-frequency and local details; (3) demonstrating excellent generalization on irregular mask models and common datasets such as Places2, Celeba-HQ, FFHQ, and the random irregular mask dataset Pconv from NVIDIA. [ABSTRACT FROM AUTHOR]

Chemical molecular structures are a direct and convenient means of expressing chemical knowledge, playing a vital role in academic communication. In chemistry, hand drawing is a common task for students and researchers. If we can convert hand-drawn chemical molecular structures into machine-readable formats, like SMILES encoding, computers can efficiently process and analyze these structures, significantly enhancing the efficiency of chemical research. Furthermore, with the progress of educational technology, automated grading is gaining popularity. When machines automatically recognize chemical molecular structures and assess the correctness of the drawings, it offers great convenience to teachers. We created ChemReco, a tool designed to identify chemical molecular structures involving three atoms: C, H, and O, providing convenience for chemical researchers. Currently, there are limited studies on hand-drawn chemical molecular structures. Therefore, the primary focus of this paper is constructing datasets. We propose a synthetic image method to rapidly generate images resembling hand-drawn chemical molecular structures, enhancing dataset acquisition efficiency. Regarding model selection, the hand-drawn chemical molecule structural recognition model developed in this article achieves a final recognition accuracy of 96.90%. This model employs the encoder-decoder architecture of EfficientNet + Transformer, demonstrating superior performance compared to other encoder-decoder combinations. [ABSTRACT FROM AUTHOR]

During their life cycle, buildings are subjected to damage that reduces their performance and can pose a significant threat to structural safety. This paper presents the results of research into the creation of a model for predicting damage intensity of buildings located in mining terrains. The basis for the research was a database of technical and mining impact data for 185 masonry residential buildings. The intensity of damage to buildings was negligible and ranged from 0 to 6%. The Convolutional Neural Network (CNN) methodology was used to create the model. The Support Vector Machine (SVM) methodology, which is commonly used for analysis of this type of issue, was used for comparisons. The resulting models were evaluated by comparing parameters such as accuracy, precision, recall, and F1 score. The comparisons revealed only minor differences between the models. Despite the small range of damage intensity, the models created were able to achieve prediction results of around 80%. The SVM model had better results for training set accuracy, while the CNN model achieved higher values for F1 score and average precision for the test set. The results obtained justify the adoption of the CNN methodology as effective in the context of predicting the damage intensity of masonry residential buildings located in mining terrains. [ABSTRACT FROM AUTHOR]

Deep convolutional neural networks have achieved remarkable progress on computer vision tasks over last years. In this paper, we proposed a dynamic training slimming with feature sparsity based on structured pruning, named DTS, for efficient and automatic channel pruning. Unlike other existing pruning methods, which require manual intervention for pruning settings for each layer, DTS can design suitable architecture width for target datasets and deployment resources by automated pruning. The proposed method can be deployed to modern CNNs and the experimental results on CIFAR, ImageNet and PASCAL VOC benchmark datasets demonstrate the effectiveness of the proposed method, which significantly exceeds the other schemes. [ABSTRACT FROM AUTHOR]

Alzheimer's disease (AD), the predominant form of dementia, is a growing global challenge, emphasizing the urgent need for accurate and early diagnosis. Current clinical diagnoses rely on radiologist expert interpretation, which is prone to human error. Deep learning has thus far shown promise for early AD diagnosis. However, existing methods often overlook focal structural atrophy critical for enhanced understanding of the cerebral cortex neurodegeneration. This paper proposes a deep learning framework that includes a novel structure-focused neurodegeneration CNN architecture named SNeurodCNN and an image brightness enhancement preprocessor using gamma correction. The SNeurodCNN architecture takes as input the focal structural atrophy features resulting from segmentation of brain structures captured through magnetic resonance imaging (MRI). As a result, the architecture considers only necessary CNN components, which comprises of two downsampling convolutional blocks and two fully connected layers, for achieving the desired classification task, and utilises regularisation techniques to regularise learnable parameters. Leveraging mid-sagittal and para-sagittal brain image viewpoints from the Alzheimer's disease neuroimaging initiative (ADNI) dataset, our framework demonstrated exceptional performance. The para-sagittal viewpoint achieved 97.8% accuracy, 97.0% specificity, and 98.5% sensitivity, while the mid-sagittal viewpoint offered deeper insights with 98.1% accuracy, 97.2% specificity, and 99.0% sensitivity. Model analysis revealed the ability of SNeurodCNN to capture the structural dynamics of mild cognitive impairment (MCI) and AD in the frontal lobe, occipital lobe, cerebellum, temporal, and parietal lobe, suggesting its potential as a brain structural change digi-biomarker for early AD diagnosis. This work can be reproduced using code we made available on GitHub. [ABSTRACT FROM AUTHOR]

Automatic pavement crack detection is a critical technique in the intelligent transportation system, which can effectively replace the person with a machine to detect the pavement crack automatically. This task is excellently challenging due to the tiny texture and the various shapes of each crack object. Previous crack detection networks mainly aim to perform complicated multi-scale feature fusion to learn the semantic information of cracks. However, the typical symmetrical networks with high-to-low resolution are undesirable to extract detailed crack texture information, and these existing methods cannot effectively deal with the issue of the various shapes.This paper proposes a novel end-to-end U-shaped convolutional neural network, termed CrackUNet, for the pavement crack detection task. To extract the information of the tiny texture, we design a novel network-in-network structure, which can enlarge the receptive field and obtain multi-scale features by putting a sub-network into each convolutional layer. Besides, to handle the issue of the various shapes for the crack object, we exploit the deformable convolution to capture contextual information of each crack, which can further improve the performance of crack detection. We train and evaluate the proposed CrackUNet on three public pavement crack datasets. The quantitative experimental results illustrate that our network outperforms the current state-of-the-art methods with almost the same efficiency. Specifically, the precision, recall, and F1-score of the CrackUNet are approximately 92.27%, 93.99%, and 92.94%, respectively. [ABSTRACT FROM AUTHOR]

In the paper, a vision-based vehicle identification system is proposed for autonomous intelligent car driving. The accurate detection of obstacles (vehicles) during intelligent car driving allows avoiding crashes, preventing accidents, saving people's lives and reducing harm. The vehicle detection system, which uses low-quality images captured by a monocular video camera mounted at the front of the car, is based on convolutional neural networks (CNN). The CNN can extract global features of the images using convolutional layers and achieves more accurate, and faithful contours of vehicles. The CNN structure proposed in the paper provides high-accuracy detection of vehicle images. The experiments that have been performed using GTI dataset demonstrate that the CNN-based vehicle detection system achieves very accurate results and is more robust to different variations of images. [ABSTRACT FROM AUTHOR]

In the case of large differences in working conditions and large noise effects, how to maintain the similarity of the same fault type is a major difficulty in the rolling bearing fault diagnosis. In addition, variable working conditions will cause the signal take place the local modulation, which makes the signal more susceptible to noise. Most of classification algorithms are difficult to eliminate the influence of variable working conditions on diagnostic results. To solve the problem, a fault diagnosis method is proposed which takes into account the change of speed and load in this paper. The method first applies the synchronous compression wavelet transformation to pre-process the data to reduce the effect of noise on feature extraction. Then, the Triplet loss function is improved and combined with the Hard Samples Mining theory to proposes a new loss function called Quadru-Hard loss to solve the problem of difficult classification under variable working conditions. Based on the experimental analysis of two sets of bearing fault data under variable speeds and variable loads, the results show that the method has a highly accuracy in fault diagnosis under two variable conditions. [ABSTRACT FROM AUTHOR]

The vibration signals of rotating machinery usually contain various natural oscillation modes, exhibiting multi-scale features. This paper proposes a Multi-Branch one-dimensional deep Convolutional Neural Network model (MBCNN) that can extract multi-scale features from raw data hierarchically, thereby improving the diagnostic accuracy of gearbox faults in noisy environments. Meanwhile, the algorithms for multi-branch generation and algorithms of the convolution and pooling for each branch are deducted. The MBCNN integrates multiple branches with interrelated convolution kernels of different widths, and each branch can extract the high-level features of the signal. The network parameters of each branch are adjusted by the loss function, which makes the features of the branches complementary. Through the design of MBCNN, the local, global, deep layer and comprehensive information can be obtained from the raw data. On the widely used Case Western Reserve University Bearing Dataset, this paper conducted a performance comparison between the proposed MBCNN and other baselines including the shallow learning methods, 1D-CNN, and multi-scale feature learning methods. Moreover, our gearbox dataset was conducted on a fault diagnosis platform, and a series of experiments were conducted to verify the effectiveness and superiority of the MBCNN. The results indicate that the MBCNN can identify the faults in the gearbox with an accuracy of higher than 92%, and the average validation time per sample is less than 3.2 ms. In a noisy environment, the diagnostic accuracy can reach 90%. The proposed MBCNN provides an effective and intelligent detection method to identify the faults of rotating machinery in the manufacturing processes. [ABSTRACT FROM AUTHOR]

Transformer models are evolving rapidly in standard natural language processing tasks; however, their application is drastically proliferating in computer vision (CV) as well. Transformers are either replacing convolution networks or being used in conjunction with them. This paper aims to differentiate the design of convolutional neural networks (CNNs) built models and models based on transformer, particularly in the domain of object detection. CNNs are designed to capture local spatial patterns through convolutional layers, which is well suited for tasks that involve understanding visual hierarchies and features. However, transformers bring a new paradigm to CV by leveraging self-attention mechanisms, which allows to capture both local and global context in images. Here, we target the various aspects such as basic level of understanding, comparative study, application of attention model, and highlighting tremendous growth along with delivering efficiency are presented effectively for object detection task. The main emphasis of this work is to offer basic understanding of architectures for object detection task and motivates to adopt the same in computer vision tasks. In addition, this paper highlights the evolution of transformer-based models in object detection and their growing importance in the field of computer vision, we also identified the open research direction in the same field. [ABSTRACT FROM AUTHOR]

This paper focuses on generating collective behavior of a robotic swarm using an attention agent. The selective attention mechanism enables an agent to cope with environmental variations which are irrelevant to the task. This paper applies attention mechanisms to a robotic swarm for enhancing system-level properties, such as flexibility or scalability. To train an attention agent, evolutionary computations become a promising method, because a controller structure is not restricted by a gradient-based method. Therefore, this paper employs a deep neuroevolution approach to generating collective behavior in a robotic swarm. The experiments are conducted by computer simulations that consist of the Unity 3D game engine. The performance of the attention agent is compared with the convolutional neural network approach. The experimental results showed that the attention agent obtained generalization abilities in a robotic swarm similar to single-agent problems. [ABSTRACT FROM AUTHOR]

With the increasing attention to the state and role of people in intelligent manufacturing, there is a strong demand for human-cyber-physical systems (HCPS) that focus on human-robot interaction. The existing intelligent manufacturing system cannot satisfy efficient human-robot collaborative work. However, unlike machines equipped with sensors, human characteristic information is difficult to be perceived and digitized instantly. In view of the high complexity and uncertainty of the human body, this paper proposes a framework for building a human digital twin (HDT) model based on multimodal data and expounds on the key technologies. Data acquisition system is built to dynamically acquire and update the body state data and physiological data of the human body and realize the digital expression of multi-source heterogeneous human body information. A bidirectional long short-term memory and convolutional neural network (BiLSTM-CNN) based network is devised to fuse multimodal human data and extract the spatiotemporal features, and the human locomotion mode identification is taken as an application case. A series of optimization experiments are carried out to improve the performance of the proposed BiLSTM-CNN-based network model. The proposed model is compared with traditional locomotion mode identification models. The experimental results proved the superiority of the HDT framework for human locomotion mode identification. [ABSTRACT FROM AUTHOR]

Multi-label image classification is a fundamental and practical task, which aims to assign multiple possible labels to an image. In recent years, many deep convolutional neural network (CNN) based approaches have been proposed which model label correlations to discover semantics of labels and learn semantic representations of images. This paper advances this research direction by improving both the modeling of label correlations and the learning of semantic representations. On the one hand, besides the local semantics of each label, we propose to further explore global semantics shared by multiple labels. On the other hand, existing approaches mainly learn the semantic representations at the last convolutional layer of a CNN. But it has been noted that the image representations of different layers of CNN capture different levels or scales of features and have different discriminative abilities. We thus propose to learn semantic representations at multiple convolutional layers. To this end, this paper designs a Multi-layered Semantic Representation Network (MSRN) which discovers both local and global semantics of labels through modeling label correlations and utilizes the label semantics to guide the semantic representations learning at multiple layers through an attention mechanism. Extensive experiments on five benchmark datasets including VOC2007, VOC2012, MS-COCO, NUS-WIDE, and Apparel show a competitive performance of the proposed MSRN against state-of-the-art models. [ABSTRACT FROM AUTHOR]

While many fast Fourier transform (FFT) architectures have been presented for computing real-valued FFT (RFFT), which of these architectures is best suited for low-throughput applications such as bio- medical signals which are typically sampled between 256 Hz and 1 kHz remains unclear. This paper implements and compares throughput, resources, and energy consumption of three different hardware architectures for real-valued FFT algorithms using Xilinx Ultra96-V2 FPGA development board. The RFFT architectures exploit the conjugate symmetry property of the real signals, thereby eliminating about half of the computations compared to a complex FFT. The three FFT architectures investigated in this paper include: single processing element (SPE), pipelined, and in-place. It is shown that, for a 256-point RFFT, using FPGA, the in-place architectures require the least device resources when compared to the pipelined architectures, while the throughput of the pipelined architectures is approximately 8 times that of the in-place architecture. These RFFT architectures are then used to generate feature vectors for a machine-learning based epileptic seizure prediction system. The seizure prediction system using the various RFFT architectures are then realized in Xilinx Ultra96-V2 FPGA development board and the power consumption values of the overall system using these architectures are compared. It is shown that the pipelined implementation of the feature extraction core results in ≈ 30 % reduction in power consumption of the entire system than the in-place implementation for the same target clock frequency, as the pipelined architecture has a higher throughput and hence is idle for majority of the computation time. [ABSTRACT FROM AUTHOR]

This research paper presents a comprehensive investigation into the utilization of color image processing technologies and deep learning algorithms in the development of a robot vision system specifically designed for 8-ball billiards. The sport of billiards, with its various games and ball arrangements, presents unique challenges for robotic vision systems. The proposed methodology addresses these challenges through two main components: object detection and ball pattern recognition. Initially, a robust algorithm is employed to detect the billiard balls using color space transformation and thresholding techniques. This is followed by determining the position of the billiard table through strategic cropping and isolation of the primary table area. The crucial phase involves the intricate task of recognizing ball patterns to differentiate between solid and striped balls. To achieve this, a modified convolutional neural network is utilized, leveraging the Xception network optimized by an innovative algorithm known as the Improved Chaos African Vulture Optimization (ICAVO) algorithm. The ICAVO algorithm enhances the Xception network's performance by efficiently exploring the solution space and avoiding local optima. The results of this study demonstrate a significant enhancement in recognition accuracy, with the Xception/ICAVO model achieving remarkable recognition rates for both solid and striped balls. This paves the way for the development of more sophisticated and efficient billiards robots. The implications of this research extend beyond 8-ball billiards, highlighting the potential for advanced robotic vision systems in various applications. The successful integration of color image processing, deep learning, and optimization algorithms shows the effectiveness of the proposed methodology. This research has far-reaching implications that go beyond just billiards. The cutting-edge robotic vision technology can be utilized for detecting and tracking objects in different sectors, transforming industrial automation and surveillance setups. By combining color image processing, deep learning, and optimization algorithms, the system proves its effectiveness and flexibility. The innovative approach sets the stage for creating advanced and productive robotic vision systems in various industries. [ABSTRACT FROM AUTHOR]

Deep learning has emerged as a highly effective and precise method for classifying images. The presence of plant diseases poses a significant threat to food security. However, accurately identifying these diseases in plants is challenging due to limited infrastructure and techniques. Fortunately, the recent advancements in deep learning within the field of computer vision have opened up new possibilities for diagnosing plant pathology. Detecting plant diseases at an early stage is crucial, and this research paper proposes a deep convolutional neural network model that can rapidly and accurately identify plant diseases. Given the minimal variation in image texture and color, deep learning techniques are essential for robust recognition. In this study, we introduce a deep, explainable neural architecture specifically designed for recognizing plant diseases. Fine-tuned deep convolutional neural network is designed by freezing the layers and adjusting the weights of learnable layers. By extracting deep features from a down sampled feature map of a fine-tuned neural network, we are able to classify these features using a customized K-Nearest Neighbors Algorithm. To train and validate our model, we utilize the largest standard plant village dataset, which consists of 38 classes. To evaluate the performance of our proposed system, we estimate specificity, sensitivity, accuracy, and AUC. The results demonstrate that our system achieves an impressive maximum validation accuracy of 99.95% and an AUC of 1, making it the most ideal and highest-performing approach compared to current state-of-the-art deep learning methods for automatically identifying plant diseases. [ABSTRACT FROM AUTHOR]

Over the past decades, the software industry has expanded to include all industries. Since stakeholders tend to use it to get their work done, software houses seek to estimate the cost of the software, which includes calculating the effort, time, and resources required. Although many researchers have worked to estimate it, the prediction accuracy results are still inaccurate and unstable. Estimating it requires a lot of effort. Therefore, there is an urgent need for modern techniques that contribute to cost estimation. This paper seeks to present a model based on deep learning and machine learning techniques by combining convolutional neural networks (CNN) and the particle swarm algorithm (PSO) in the context of time series forecasting, which enables feature extraction and automatic tuning of hyperparameters, which reduces the manual effort of selecting parameters and contributes to fine-tuning. The use of PSO also enhances the robustness and generalization ability of the CNN model and its iterative nature allows for efficient discovery of hyperparameter similarity. The model was trained and tested on 13 different benchmark datasets and evaluated through six metrics: mean absolute error (MAE), mean square error (MSE), mean magnitude relative error (MMRE), root mean square error (RMSE), median magnitude relative error (MdMRE), and prediction accuracy (PRED). Comparative results reveal that the performance of the proposed model is better than other methods for all datasets and evaluation criteria. The results were very promising for predicting software cost estimation. [ABSTRACT FROM AUTHOR]

The paper developing a resilient speech classification system for individuals with voice disorders poses a formidable challenge due to the significant variability and distortions inherent in vocal signals. This article outlines the steps to create an effective classification system for pathological speech. The first step involved applying speech enhancement processing using the minimum mean square error (MMSE) enhancer to improve voice input data quality and intelligibility. Secondly, a multi-stream approach combined various acoustic vectors based on human auditory perception, including mel-spectrogram images, mel frequency cepstral coefficients (MFCC), power normalized cepstral coefficients (PNCC), and prosodic parameters like F0, Jitter, and Shimmer. Finally, a deep machine learning incorporating both a convolutional neural network (CNN) and a bidirectional long short-term memory (BiLSTM) network was employed to process these enhanced characteristics in a multi-stream framework, resulting in a powerful classification system architecture. In our experiments, we utilized two subsets from the Massachusetts Eye and Ear Infirmary (MEEI) database, each involving distinct causes of voice disorders. The first subset consisted of voice recordings from patients with vocal nodules, paralysis, and polyps, while the second subset included recordings from patients with mild ventricular compression, A–P squeezing, and gastric reflux. The results we obtained reveal that the CNN-BiLSTM system, coupled with a robust speech analysis interface based on the multi-stream approach and enhanced by the minimum mean square error (MMSE) processing, achieved the highest accuracy rates. [ABSTRACT FROM AUTHOR]

Recently, the pandemic caused by COVID-19 is severe in the entire world. The prevention and control of crimes associated with COVID-19 are critical for controlling the pandemic. Therefore, to provide efficient and convenient intelligent legal knowledge services during the pandemic, we develop an intelligent system for legal information retrieval on the WeChat platform in this paper. The data source we used for training our system is "The typical cases of national procuratorial authorities handling crimes against the prevention and control of the new coronary pneumonia pandemic following the law", which is published online by the Supreme People's Procuratorate of the People's Republic of China. We base our system on convolutional neural network and use the semantic matching mechanism to capture inter-sentence relationship information and make a prediction. Moreover, we introduce an auxiliary learning process to help the network better distinguish the relation between two sentences. Finally, the system uses the trained model to identify the information entered by a user and responds to the user with a reference case similar to the query case and gives the reference legal gist applicable to the query case. [ABSTRACT FROM AUTHOR]

Identifying and preventing malicious network behavior is a challenge for establishing a secure network communication environment or system. Malicious activities in a network system can seriously threaten users' privacy and potentially jeopardize the entire network infrastructure and functions. Furthermore, cyber-attacks have grown in complexity and number due to the ever-evolving digital landscape of computer and network devices in recent years. Analyzing network traffic using network intrusion detection systems (NIDSs) has become an integral security measure in modern networks to identify malicious and suspicious activities. However, most intrusion detection datasets contain imbalance classes, making it difficult for most existing classifiers to achieve good performance. In this paper, we propose a double-layer feature extraction and feature fusion technique (CNN-GRU-FF), which uses a modified focal loss function instead of the traditional cross-entropy to handle the class imbalance problem in the IDS datasets. We use the NSL-KDD and UNSW-NB15 datasets to evaluate the effectiveness of the proposed model. From the research findings, it is evident our CNN-GRU-FF method obtains a detection rate of 98.22% and 99.68% using the UNSW-NB15 and NSL-KDD datasets, respectively while maintaining low false alarm rates on both datasets. We compared the proposed model's performance with seven baseline algorithms and other published methods in literature. It is evident from the performance results that our proposed method outperforms the state-of-the-art network intrusion detection methods. [ABSTRACT FROM AUTHOR]

Drowsiness detection is a critical aspect of ensuring safety in various domains, including transportation, online learning, and multimedia consumption. This research paper presents a comprehensive investigation into drowsiness detection methods, with a specific focus on utilizing convolutional neural networks (CNN) and transfer learning. Notably, the proposed study extends beyond theoretical exploration to practical application, as we have developed a user-friendly mobile application incorporating these advanced techniques. Diverse datasets are integrated to systematically evaluate the implemented model, and the results showcase its remarkable effectiveness. For both multi-class and binary classification scenarios, our drowsiness detection system achieves impressive accuracy rates ranging from 90 to 99.86%. This research not only contributes to the academic understanding of drowsiness detection but also highlights the successful implementation of such methodologies in real-world scenarios through the development of our application. [ABSTRACT FROM AUTHOR]

Background: Image denoising technology removes noise from the corrupted image by utilizing different features between image and noise. Convolutional neural network (CNN)-based algorithms have been the concern of the recent progress on diverse image restoration problems and become an efficient solution in image denoising. Objective: Although a quite number of existing CNN-based image denoising methods perform well on the simplified additive white Gaussian noise (AWGN) model, their performance often degrades severely on the real-world noisy images which are corrupted by more complicated noise. Methods: In this paper, we utilized the multi-task learning (MTL) framework to integrate multiple loss functions for collaborative training of CNN. This approach aims to improve the denoising performance of CNNs on real-world images with non-Gaussian noise. Simultaneously, to automatically optimize the weights of individual sub-tasks within the MTL framework, we incorporated a self-learning weight layer into the CNN. Results: Extensive experiments demonstrate that our approach effectively enhances the denoising performance of CNN-based image denoising algorithms on real-world images. It reduces excessive image smoothing, improves quantitative metrics, and enhances visual quality in the restored images. Conclusion: Our method shows the effectiveness of the improved performance of denoising CNNS for real-world image denoising processing. [ABSTRACT FROM AUTHOR]

It is desired to apply deep learning models (DLMs) to assist physicians in distinguishing abnormal/normal lung sounds as quickly as possible. The performance of DLMs depends on feature-related and model-related parameters heavily. In this paper, the relationship between performance and feature-related parameters of a DLM, i.e., convolutional neural network (CNN) is analyzed through experiments. ICBHI 2017 is selected as the lung sounds dataset. The sensitivity analysis of classification performance of the DLM on three parameters, i.e., the length of lung sounds frame, overlap percentage (OP) of successive frames and feature type, is performed. An augmented and balanced dataset is acquired by the way of white noise addition, time stretching and pitch shifting. The spectrogram and mel frequency cepstrum coefficients of lung sounds are used as features to the CNN, respectively. The results of training and test show that there exists significant difference on performance among various parameter combinations. The parameter OP is performance sensitive. The higher OP, the better performance. It is concluded that for fixed sampling frequency 8 kHz, frame size 128, OP 75% and spectrogram feature is optimum under which the performance is relatively better and no extra computation or storage resources are required. [ABSTRACT FROM AUTHOR]

The increasing complexity and adversity of cyber-attacks have prompted discussions in the cyber scenario for a prognosticate approach, rather than a reactionary one. In this paper, a signature-based intrusion detection system has been built based on C5 classifiers, to classify packets into normal and attack categories. Next, an anomaly-based intrusion detection was built based on the LSTM (Long-Short Term Memory) algorithm to detect anomalies. These anomalies are then fed into the signature generator to extract attributes. These attributes get uploaded into the C5 training set, aiding the ensemble model in continual learning with expanding signatures of unknown attacks. By generating signatures of unknown attacks, the self-healing attribute of the ensemble model contributes to the early detection of attacks. For the C5 classifier, the proposed model is evaluated on the UNSW-NB15 dataset, while for the LSTM model, it is evaluated on the ADFA-LD dataset. Compared to conventional models, the experimental results show better detection rates for both known and unknown attacks. The C5 classifier achieved a True Positive Rate of 97% while maintaining a false positive rate of 8%. Also, the LSTM model achieved a detection rate of 90% while retaining a 17% False Alarm Rate. As the proposed model learns, its performance in real network traffic also improves and it also eliminates human intervention when updating training data. [ABSTRACT FROM AUTHOR]

Scene text image super-resolution is an interesting and challenging task which aims to enhance the spatial resolution of low-resolution text images in the wild, and consequently improve the image visual quality and boost the performance of real-world text-related applications. However, most of previous super-resolution methods ignore the important specific characteristics of text patterns and regard scene text images as natural scene images. In this paper, a novel deep convolutional-based architecture is specifically proposed for the super-resolution of scene text images. In order to recover fine details of low-resolution characters, the proposed architecture has been carefully designed and its main specificities are three-fold: (1) the introduction of multi-scale features extraction by incorporating parallel convolutional layers in order to preserve both local and global high-frequency components that encapsulate the intricate details of characters' patterns. This strategy allows the proposed method to capture fine nuances in the visual representation of characters, enhancing the richness of extracted features. (2) the integration of skip connections through convolutional layers. This strategic design choice facilitates the seamless flow of information from lower to higher layers of the deep architecture, allowing sequential information about text patterns to be preserved more effectively. (3) the proposition of a specialized network in network-based reconstruction within our architecture to recover high-resolution text details from the collected features. Such a network paradigm minimizes information loss and enhances the proposed method's ability to discern and reconstruct fine textual details. These design elements collectively empower our super-resolution method to excel in analyzing fine text patterns for effective high-resolution reconstruction, providing a comprehensive solution for the challenging task of recovering fine details in low-resolution characters. Quantitative and qualitative evaluations on four well-known benchmarks, including the SVT, IIIT5k, IC03 and ICDAR2015-TextSR datasets, prove the efficiency of our proposal whose performance surpasses those of different state-of-the-art super-resolution methods. [ABSTRACT FROM AUTHOR]

In order to explore the application of neural networks in the intelligent protection of digital music copyrights, this paper initially investigates the key aspects of digital music-related supply chain management (SCM). Subsequently, this paper introduces a fuzzy, comprehensive evaluation-based method for authenticating digital music infringement. Furthermore, it employs a convolutional neural network to analyze digital music's source components. Within this context, an attention mechanism and an adaptive gate model are meticulously designed. Moreover, a selective adaptive cascade is implemented to optimize the structure of the multi-resolution coder–decoder. Through fuzzy authentication, it becomes feasible to accurately and reliably track and verify copyright information pertaining to digital music, thus providing an effective protective mechanism for the digital music industry. Experimental results reveal that in comparison with the SHN-4 series models, the network model embedded with the channel space (CS) attention module achieves improvements of up to 0.04 dB in terms of performance metrics for human voice and accompaniment sources. Furthermore, it outperforms these models by up to 0.19 dB in music source separation metrics. The incorporation of the adaptive attention mechanism profoundly enhances the model's performance in audio source separation tasks, benefiting both vocal and accompaniment sources. This improvement is evident in the form of a 0.44 SDR enhancement for vocal sources and a 0.18 SDR enhancement for accompaniment sources. Additionally, the proposed two-stage music separation-gate self-attention model surpasses the two-stage music separation-gate model in metrics associated with human voice and accompaniment sources. This outcome convincingly underscores the enhanced effectiveness achieved by integrating CS, self-attention, and adaptive connection in network performance, as well as the adaptive gate structure for music copyright protection. This approach heightens the model's autonomous selectivity, minimizes outliers, stabilizes waveform characteristics, and significantly contributes to the safeguarding of digital content. [ABSTRACT FROM AUTHOR]

Radar signal sorting is a vital component of electronic warfare reconnaissance, serving as the basis for identifying the source of radar signals. However, traditional radar signal sorting methods are increasingly inadequate and computationally complex in modern electromagnetic environments. To address this issue, this paper presents a novel machine-learning-based approach for radar signal sorting. Our method utilizes SemHybridNet, a Semantically Enhanced Hybrid CNN-Transformer Network, for the classification of semantic information in two-dimensional radar pulse images obtained by converting the original radar data. SemHybridNet incorporates two innovative modules: one for extracting period structure features, and the other for ensuring effective integration of local and global features. Notably, SemHybridNet adopts an end-to-end structure, eliminating the need for repetitive looping over the original sequence and reducing computational complexity. We evaluate the performance of our method through conducting comprehensive comparative experiments. The results demonstrate our method significantly outperforms the traditional methods, particularly in environments with high missing and noise pulse rates. Moreover, the ablation studies confirm the effectiveness of these two proposed modules in enhancing the performance of SemHybridNet. In conclusion, our method holds promise for enhancing electronic warfare reconnaissance capabilities and opens new avenues for future research in this field. [ABSTRACT FROM AUTHOR]