Search

Modern manufacturing is the world's largest and most automated industrial sector. The rise of Industry 4.0 technologies such as Big Data, Internet of Things (IoT) devices, and Machine Learning has enabled a better connection with machines and factory systems. Data harvesting allowed for a more seamless and comprehensive implementation of the knowledge-based decision-making process. New models that provide a competitive edge must be created by combining the Lean paradigm with the new technologies of Industry 4.0. This paper presents novel computer-based vision models for automated detection and classification of damaged packages from intact packages. In high-volume production environments, the package manual inspection process through the human eye consumes inordinate amounts of time poring over physical packages. Our proposed three different computer-based vision approaches detect damaged packages to prevent them from moving to shipping operations that would otherwise incur waste in the form of wasted operating hours, wasted resources and lost customer satisfaction. The proposed approaches were carried out on a data set consisting of package images and achieved high precision, accuracy, and recall values during the training and validation stage, with the resultant trained YOLO v7 model. [ABSTRACT FROM AUTHOR]

Traffic signs have great importance in driving safety. For the recently emerging autonomous vehicles, that can automatically detect and recognize all road inventories such as traffic signs. Firstly, in this study, a method based on a mobile mapping system (MMS) is proposed for the detection of traffic signs to establish a Turkish traffic sign dataset. Obtaining images from real traffic scenes using the MMS method enhances the reliability of the model. It is an easy method to be applied to real life in terms of both cost and suitability for mobile and autonomous systems. In this frame, YOLOv4-Tiny, one of the object detection algorithms, that is considered to be more suitable for mobile vehicles, is used to detect and recognize traffic signs. This algorithm is low operation cost and more suitable for embedded devices due to its simple neural network structure compared to other algorithms. It is also a better option for real-time detection than other approaches. For the training of the model in the suggested method, a dataset consisting partly of images taken with MMS based on realistic field measurement and partly of images obtained from open data sets was used. This training resulted in the mean average precision (mAP) value being obtained as 98.1%. The trained model was first tested on existing images and then tested in real time in a laboratory environment using a simple fixed web camera. The test results show that the suggested method can improve driving safety by detecting traffic signs quickly and accurately, especially for autonomous vehicles. Therefore, the proposed method is considered suitable for use in autonomous vehicles. [ABSTRACT FROM AUTHOR]

Saliency maps are input-resolution matrices used for visualizing local interpretations of image classifiers. Their pixel values reflect the importance of corresponding image locations for the model's decision. Despite numerous proposals on how to obtain such maps, their evaluation remains an open question. This paper presents a carefully designed experimental procedure along with a set of quantitative interpretation evaluation metrics that rely solely on the original model behavior. Previously noticed evaluation biases have been attenuated by separating locations with high and low values, considering the full saliency map resolution, and using classifiers with diverse accuracies and all the classes in the dataset. We used the proposed evaluation metrics to compare and analyze seven well-known interpretation methods. Our experiments confirm the importance of object background as well as negative saliency map pixels, and we show that the scale of their impact on the model is comparable to that of positive ones. We also demonstrate that a good class score interpretation does not necessarily imply a good probability interpretation. DeepLIFT and LRP- ϵ methods proved most successful altogether, while Grad-CAM and Ablation-CAM performed very poorly, even in the detection of positive relevance. The retention of positive values alone in the latter two methods was responsible for the inaccurate detection of irrelevant locations as well. [ABSTRACT FROM AUTHOR]

Agriculture is important for the economy of any country, and India is considered to be an agricultural country. One of the primary goals of agriculture is to produce disease-free crops. Since ancient times, farmers and other planting specialists have had to contend with a variety of problems and current agricultural constraints, such as widespread cotton diseases. There is a great need for a rapid, efficient, economical, and reliable approach to diagnosing cotton infection in the agri-informatics area, as severe cotton disease may result in the loss of grain crops. This paper presents an advanced method that automates the detection and classification of diseased cotton leaves and plants through deep learning techniques applied to images. To address the challenge of supervised image classification, we employ a bagging ensemble technique consisting of five transfer learning models: InceptionV3, InceptionResNetV2, VGG16, MobileNet, and Xception. This ensemble approach was adopted to significantly improve the performance of each individual mode. The ETL-NET framework we introduced was thoroughly evaluated using two publicly accessible datasets. Specifically, it achieved an impressive accuracy rate of 99.48% and a sensitivity rate of 99% when applied to binary datasets. Additionally, on the multi-class dataset, the framework achieved an accuracy rate of 98.52% and a sensitivity rate of 99%. Our method outperformed the state-of-the-art techniques and displayed comparatively better results. Remarkably, our approach demonstrated even higher performance than widely used ensemble techniques, generally considered benchmarks in the field. [ABSTRACT FROM AUTHOR]

Glaucoma is one of the leading causes of visual impairment worldwide. If diagnosed too late, the disease can irreversibly cause severe damage to the optic nerve, resulting in permanent loss of central vision and blindness. Therefore, early diagnosis of the disease is critical. Recent advancements in machine learning techniques have greatly aided ophthalmologists in timely and efficient diagnosis through the use of automated systems. Training the machine learning models with the most informative features can significantly enhance their performance. However, selecting the most informative feature subset is a real challenge because there are 2n potential feature subsets for a dataset with n features, and the conventional feature selection techniques are also not very efficient. Thus, extracting relevant features from medical images and selecting the most informative is a challenging task. Additionally, a considerable field of study has evolved around the discovery and selection of highly influential features (characteristics) from a large number of features. Through the inclusion of the most informative features, this method has the potential to improve machine learning classifiers by enhancing their classification performance, reducing training and testing time, and lowering system diagnostic costs by incorporating the most informative features. This work aims in the same direction to propose a unique, novel, and highly efficient feature selection (FS) approach using the Whale Optimization Algorithm (WOA), the Grey Wolf Optimization Algorithm (GWO), and a hybridized version of these two metaheuristics. To the best of our knowledge, the use of these two algorithms and their amalgamated version for FS in human disease prediction, particularly glaucoma prediction, has been rare in the past. The objective is to create a highly influential subset of characteristics using this approach. The suggested FS strategy seeks to maximize classification accuracy while reducing the total number of characteristics used. We evaluated the efficacy of the proposed approach in classifying eye-related glaucoma illnesses. In this study, we aim to assist professionals in identifying glaucoma by utilizing a proposed clinical decision support system that integrates image processing, soft-computing algorithms, and machine learning, and validates it on benchmark fundus images. Initially, we extract 65 features from the 646 retinal fundus images in the ORIGA benchmark dataset, from which a subset of features is created. For two-class classification, different machine learning classifiers receive the elected features. Employing 5-fold and 10-fold stratified cross-validation has enhanced the generalized performance of the proposed model. We assess performance using several well-established statistical criteria. The tests show that the suggested computer-aided diagnosis (CAD) model has an F1-score of 97.50%, an accuracy score of 96.50%, a precision score of 97%, a sensitivity score of 98.10%, a specificity score of 93.30%, and an AUC score of 94.2% on the ORIGA dataset. To demonstrate its excellence, we compared the suggested approach's performance with other current state-of-the-art models. The suggested approach shows promising results in predicting glaucoma, potentially aiding in the early diagnosis and treatment of the disease. Furthermore, real-time applications showcase the proposed approach's suitability, enabling its deployment in areas lacking expert medical practitioners. Overburdened expert ophthalmologists can use this approach as a second opinion, as it requires very little time for processing the retinal fundus images. The proposed model can also aid, after incorporating required modifications, in making clinical decisions for various diseases like lung infection and, diabetic retinopathy. [ABSTRACT FROM AUTHOR]

We propose a technique for first‐order gradient‐based optimization of stochastic objective functions called Nesterov‐accelerated adaptive moment assessment, which makes use of dynamic evaluations of lower‐order moments. The adaptive moment assessment and the Nesterov acceleration gradient are combined. Consequently, it has perks, and this technique is convenient to use, numerically economical, memory‐light, and very well‐suited for challenges with massive amounts of information and characteristics. Additionally, we investigate the algorithm's convergence characteristics and propose a conservative constraint on the convergence rate. Finally, we employ this technique for the detection and classification of safety helmets. [ABSTRACT FROM AUTHOR]

To tackle challenges in road multi‐object detection, such as object occlusion, small object detection, and multi‐scale object detection difficulties, a new YOLOv8n‐RSFM structure is proposed. The key improvement of this structure lies in the introduction of the transformer decoder head, which optimizes the matching between the ground truth and predicted boxes, thereby effectively addressing issues of object overlap and multi‐scale detection. Additionally, a small object detection layer is incorporated to retain crucial information beneficial for detecting small objects, significantly improving the detection accuracy for small targets. To enhance learning capacity and reduce redundant computations, the FasterNet backbone is employed to replace CSPDarknet53, thus accelerating the training process. Finally, the INNER‐MPDIoU loss function is introduced to replace the original algorithm's complete IoU to accelerate the convergence and obtain more accurate regression results. A series of experiments were conducted on different datasets. The experimental results show that the proposed model YOLOv8N‐RSFM outperforms the original model YOLOv8n in small target detection. On the VisDrone, TinyPerson, and VSCrowd datasets, the mean accuracy percentage improved by 7.9%, 12.3%, and 4.5%, respectively. [ABSTRACT FROM AUTHOR]

Due to the powerful ability in capturing the global information, transformer has become an alternative architecture of CNNs for hyperspectral image classification. However, general transformer mainly considers the global spectral information while ignores the multiscale spatial information of the hyperspectral image. In this paper, we propose a multiscale spectral–spatial convolutional transformer (MultiFormer) for hyperspectral image classification. First, the developed method utilizes multiscale spatial patches as tokens to formulate the spatial transformer and generates multiscale spatial representation of each band in each pixel. Second, the spatial representation of all the bands in a given pixel are utilized as tokens to formulate the spectral transformer and generate the multiscale spectral–spatial representation of each pixel. Besides, a modified spectral–spatial CAF module is constructed in the MultiFormer to fuse cross‐layer spectral and spatial information. Therefore, the proposed MultiFormer can capture the multiscale spectral–spatial information and provide better performance than most of other architectures for hyperspectral image classification. Experiments are conducted over commonly used real‐world datasets and the comparison results show the superiority of the proposed method. [ABSTRACT FROM AUTHOR]

This research investigates advanced approaches in medical image analysis, specifically focusing on segmentation and classification techniques, as well as their integration into multi‐task architectures for lung infections. This research begins by explaining key architectural models used in segmentation and classification tasks. The study extends to the enhancement of these architectures through attention modules and conditional random fields. Relevant datasets and evaluation metrics, incorporating discussions on loss functions are also reviewed. This review encompasses recent advancements in single‐task and multi‐task models, highlighting innovations in semi‐supervised, self‐supervised, few‐shot, and zero‐shot learning techniques. Empirical analysis is conducted on both single‐task and multi‐task architectures, predominantly utilizing the U‐Net framework, and is applied across multiple datasets for segmentation and classification tasks. Results demonstrate the effectiveness of these models and provide insights into the strengths and limitations of different approaches. This research contributes to improved detection and diagnosis of lung infections by offering a comprehensive overview of current methodologies and their practical applications. [ABSTRACT FROM AUTHOR]

License plate recognition is crucial in Intelligent Transportation Systems (ITS) for vehicle management, traffic monitoring, and security inspection. In highway scenarios, this task faces challenges such as diversity, blurriness, occlusion, and illumination variation of license plates. This article explores Recurrent Neural Networks based on Connectionist Temporal Classification (RNN‐CTC) for license plate recognition in challenging highway conditions. Four neural network models: ResNet50, ResNeXt, InceptionV3, and SENet, all combined with RNN‐CTC are comparatively evaluated. Furthermore, a novel architecture named ResNet50 Deep Fusion Network using Connectionist Temporal Classification (ResNet50‐DFN‐CTC) is proposed. Comparative and ablation experiments are conducted using the Highway License Plate Dataset of Southeast University (HLPD‐SU). Results demonstrate the superior performance of ResNet50‐DFN‐CTC in challenging highway conditions, achieving 93.158% accuracy with a processing time of 7.91 ms, outperforming other tested models. This research contributes to advancing license plate recognition technology for real‐world highway applications under adverse conditions. [ABSTRACT FROM AUTHOR]

Melanoma, a highly prevalent and lethal form of skin cancer, has a significant impact globally. The chances of recovery for melanoma patients substantially improve with early detection. Currently, deep learning (DL) methods are gaining popularity in assisting with the early identification of melanoma. Despite their high performance, relying solely on an image classifier undermines the credibility of the application and makes it difficult to understand the rationale behind the model's predictions highlighting the need for Explainable AI (XAI). This study provides a survey on skin cancer identification using DL techniques utilized in studies from 2017 to 2024. Compared to existing survey studies, the authors address the latest related studies covering several public skin cancer image datasets and focusing on segmentation, classification based on convolutional neural networks and vision transformers, and explainability. The analysis and the comparisons of the existing studies will be beneficial for the researchers and developers in this area, to identify the suitable techniques to be used for automated skin cancer image classification. Thereby, the survey findings can be used to implement support applications advancing the skin cancer diagnosis process. [ABSTRACT FROM AUTHOR]

Glaucoma is an eye disease that damages the optic nerve as a result of vision loss, it is the leading cause of blindness worldwide. Due to the time‐consuming, inaccurate, and manual nature of traditional methods, automation in glaucoma detection is important. This paper proposes an explainable artificial intelligence (XAI) based model for automatic glaucoma detection using pre‐trained convolutional neural networks (PCNNs) and machine learning classifiers (MLCs). PCNNs are used as feature extractors to obtain deep features that can capture the important visual patterns and characteristics from fundus images. Using extracted features MLCs then classify glaucoma and healthy images. An empirical selection of the CNN and MLC parameters has been made in the performance evaluation. In this work, a total of 1,865 healthy and 1,590 glaucoma images from different fundus datasets were used. The results on the ACRIMA dataset show an accuracy, precision, and recall of 98.03%, 97.61%, and 99%, respectively. Explainable artificial intelligence aims to create a model to increase the user's trust in the model's decision‐making process in a transparent and interpretable manner. An assessment of image misclassification has been carried out to facilitate future investigations. [ABSTRACT FROM AUTHOR]

Advances in Unmanned Aerial Vehicles (UAVs), otherwise recognized as drones, have tremendous promise in improving the wide-ranging applications of the Internet of Things (IoT). UAV image classification using deep learning (DL) is an amalgamation to modernize data analysis, collection, and decision-making in a variety of sectors. IoT devices collect information in real time, while remote sensing captures data afar without direct contact. UAVs equipped with sensors offer high-quality images for classification tasks. DL techniques, especially the convolutional neural networks (CNNs), analyze data streams, extracting complicated features for the accurate classification of objects or environmental features. This synergy enables applications including urban planning and precision agriculture, fostering smarter disaster response, decision support systems, and efficient resource management. This paper introduces a novel Pyramid Channel-based Feature Attention Network with an Ensemble Learning-based UAV Image Classification (PCFAN-ELUAVIC) technique in an assisted remote sensing environment. The PCFAN-ELUAVIC technique begins with the contrast enhancement of the UAV images using the CLAHE technique. Following that, the feature vectors are derived by the use of the PCFAN model. Meanwhile, the hyperparameter tuning procedure is executed by the inclusion of a vortex search algorithm (VSA). For image classification, the PCFAN-ELUAVIC technique comprises an ensemble of three classifiers like long short-term memory (LSTM), graph convolutional networks (GCNs), and Hermite neural network (HNN). To exhibit the improved detection results of the PCFAN-ELUAVIC system, an extensive range of experiments are carried out. The experimental values confirmed the enhanced performance of the PCFAN-ELUAVIC model when compared to other techniques. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The recent emergence of the hybrid quantum-classical neural network (HQCNN) architecture has garnered considerable attention because of the potential advantages associated with integrating quantum principles to enhance various facets of machine learning algorithms and computations. However, the current investigated serial structure of HQCNN, wherein information sequentially passes from one network to another, often imposes limitations on the trainability and expressivity of the network. In this study, we introduce a novel architecture termed parallel proportional fusion of spiking and quantum neural networks (PPF-SQNN). The dataset information is simultaneously fed into both the spiking neural network and the variational quantum circuits, with the outputs amalgamated in proportion to their individual contributions. We systematically assess the impact of diverse PPF-SQNN parameters on network performance for image classification, aiming to identify the optimal configuration. On three datasets for image classification tasks, the final classification accuracy reached 98.2%, 99.198%, and 97.921%, respectively, with loss values all below 0.2, outperforming the compared serial networks. In noise testing, it also demonstrates good classification performance even under noise intensities of 0.9 Gaussian and uniform noise. This study introduces a novel and effective amalgamation approach for HQCNN, laying the groundwork for the advancement and application of quantum advantages in artificial intelligence computations. [ABSTRACT FROM AUTHOR]

Alzheimer's disease (AD) is a progressive neurological disorder characterized by the gradual deterioration of cognitive functions, leading to dementia and significantly impacting the quality of life for millions of people worldwide. Early and accurate diagnosis is crucial for the effective management and treatment of this debilitating condition. This study introduces a novel framework based on Spectral Graph Convolutional Neural Networks (SGCNN) for diagnosing AD and categorizing multiple diseases through the analysis of functional changes in brain structures captured via magnetic resonance imaging (MRI). To assess the effectiveness of our approach, we systematically analyze structural modifications to the SGCNN model through comprehensive ablation studies. The performance of various Convolutional Neural Networks (CNNs) is also evaluated, including SGCNN variants, Base CNN, Lean CNN, and Deep CNN. We begin with the original SGCNN model, which serves as our baseline and achieves a commendable classification accuracy of 93%. In our investigation, we perform two distinct ablation studies on the SGCNN model to examine how specific structural changes impact its performance. The results reveal that Ablation Model 1 significantly enhances accuracy, achieving an impressive 95%, while Ablation Model 2 maintains the baseline accuracy of 93%. Additionally, the Base CNN model demonstrates strong performance with a classification accuracy of 93%, whereas both the Lean CNN and Deep CNN models achieve 94% accuracy, indicating their competitive capabilities. To validate the models' effectiveness, we utilize multiple evaluation metrics, including accuracy, precision, recall, and F1-score, ensuring a thorough assessment of their performance. Our findings underscore that Ablation Model 1 (SGCNN Model 1) delivers the highest predictive accuracy among the tested models, highlighting its potential as a robust approach for Alzheimer's image classification. Ultimately, this research aims to facilitate early diagnosis and treatment of AD, contributing to improved patient outcomes and advancing the field of neurodegenerative disease diagnosis. [ABSTRACT FROM AUTHOR]

This study aims to explore an autoencoder-based method for generating brain MRI images of patients with Autism Spectrum Disorder (ASD) and non-ASD individuals, and to discriminate ASD based on the generated images. Initially, we introduce the research background of ASD and related work, as well as the application of deep learning in the field of medical imaging. Subsequently, we detail the architecture and training process of the proposed autoencoder model, and present the results of generating MRI images for ASD and non-ASD patients. Following this, we designed an ASD classifier based on the generated images and elucidated its structure and training methods. Finally, through analysis and discussion of experimental results, we validated the effectiveness of the proposed method and explored future research directions and potential clinical applications. This research offers new insights and methodologies for addressing challenges in ASD studies using deep learning technology, potentially contributing to the automated diagnosis and research of ASD. [ABSTRACT FROM AUTHOR]

The classification of ablation images holds significant practical value in thermal protection structures, as it enables the assessment of heat and corrosion resistance of composites. This paper proposes an image‐based deep learning framework to identify the surface texture of carbon/phenolic composites ablative images. First, ablation experiments and collection of surface texture images of carbon/phenolic composites under different thermal environments were conducted in an electric arc wind tunnel. Then, a deep learning model based on a convolutional neural network (CNN) is developed for ablative image classification. The pre‐trained network is ultimately employed as the input for transfer learning. The network's feature extraction layer is trained using the ImageNet dataset, while the global average pooling addresses specific classification tasks. The test results demonstrate that the proposed method effectively classifies the relatively small surface texture dataset, enhances the classification performance of ablative surface texture with an accuracy of up to 97.6%, and exhibits robustness and generalization capabilities. Highlights: The paper proposes a new deep learning classification method for ablative images.A model highly sensitive to small and weak features is built.Transfer learning and data enhancement techniques are introduced into classification. [ABSTRACT FROM AUTHOR]

The use of social media data for disaster-type identification has been turning progressively important in recent years. With the extensive dependency on social networking sites, people can share real-time information and updates about disasters, making it a valuable source of information for disaster management organizations. The use of natural language processing (NLP) and computer vision techniques can help process and examine large amounts of social media data to gain valuable insights into the nature and extent of a disaster. In this study, NLP, and convolutional neural networks (CNN) models were applied to social media data for disaster-type recognition. The language models used were BERT-Base-Uncased, DistilBERT-Base-Uncased, Twitter-RoBERTa-Base, and FinBERT. Two convolutional neural network (CNN) models, Inception v3 and DenseNet were also applied. The models were evaluated on the CrisisMMD dataset. The outcome proved that the language models achieved a uniform accuracy of 94% across disaster-related tweet classification tasks, while DistilBERT-Base-Uncased demonstrated the fastest training and testing time which is important for prompt response systems. In terms of the CNN models, DenseNet outperformed Inception v3 just by a small margin of 1 or 2% in terms of accuracy, recall, precision, and F1 score. This entails that the DistilBERT-Base-Uncased and DenseNet model has the potential to be better suited for disaster-type recognition using social media data in terms of accuracy and time. [ABSTRACT FROM AUTHOR]

Coherent anti‐Stokes Raman scattering (CARS) microscopy is a powerful label‐free imaging technique that leverages biomolecular vibrations and is widely used in different fields. However, its intrinsic non‐resonant background (NRB) can distort Raman signals and compromise spectral fidelity. Conventional data analysis methods for CARS encounter a bottleneck in achieving high accuracy. Furthermore, CARS requires balancing imaging speed against image quality. In recent years, endeavors in deep learning have effectively overcome these obstacles, advancing the development of CARS. This review highlights the research that applies deep learning to mitigate NRB, classify CARS data for disease identification, and denoise images. Each approach is delineated in terms of network architecture, training data, and loss functions. Finally, the challenges in this field is discussed and using the latest deep learning advancement is suggested to enhance the reliability and efficiency of CARS microscopy. [ABSTRACT FROM AUTHOR]

Germinal matrix hemorrhage (GMH) is a critical condition affecting premature infants, commonly diagnosed through cranial ultrasound imaging. This study presents an advanced deep learning approach for automated GMH grading using the YOLOv8 model. By analyzing a dataset of 586 infants, we classified ultrasound images into five distinct categories: Normal, Grade 1, Grade 2, Grade 3, and Grade 4. Utilizing transfer learning and data augmentation techniques, the YOLOv8 model achieved exceptional performance, with a mean average precision (mAP50) of 0.979 and a mAP50-95 of 0.724. These results indicate that the YOLOv8 model can significantly enhance the accuracy and efficiency of GMH diagnosis, providing a valuable tool to support radiologists in clinical settings. [ABSTRACT FROM AUTHOR]

To improve users' satisfaction in purchasing clothes on e-commerce platforms shopping websites currently use manual input text or search pictures to help consumers find clothes. However it is quite difficult for consumers to find satisfactory clothing in a short time when the information of clothing is not detailed. To quickly find clothing that meets the needs of consumers shopping websites shall classify clothes according to user' preferences and styles and label each item of clothes. However for huge garment image data manual annotation classification methods require a lot of time and labour and are susceptible to human subjective factors. Therefore it is necessary to study the method of quick picture classification using computer networks. The algorithm based on deep learning can extract useful features from large-scale clothing image data and then classify them with much higher accuracy and speed than traditional manual labeling and classification methods. This shortens consumers' search time and narrows the search scope thus providing consumers with more accurate results. To address the problem of inaccurate classification of pant silhouette recognition and classification models the Vision Transformer model with self-attention mechanism was adopted to realize the classification of pant silhouette images. Firstly by adding self-attention mechanism and enhancing useful feature the interference problem of image background and other irrelevant information on profile recognition was solved. Secondly the transfer learning method was used to train and verify the profiles of four types of pants including wide-leg pants bell-bottom pants tights and Harem pants. Finally the improved Vision Transformer model was compared with the traditional CNN model to verify the model effect. In this paper the CNN structure was introduced into the input layer of the Vision Transformer model Dropout was replaced by DropPath method at the coding layer and the output layer was set to a fully connected layer. Then transfer learning was used to train the improved Vision Transformer model on the enhanced data set of this study. Experimental results show that the method of data enhancement can effectively improve model generalization ability. The average recognition rate of the test set is increased from 93. 99% before data enhancement to 95. 48% after data enhancement. After comparison of four models the recognition accuracy rate of the improved Vision Transformer model in the validation set reaches 97. 72% being 10% higher than the MobileNetV2 model. The model in this paper has a good ability to identify the silhouette of trousers and has a high application value in the classification and recognition of clothing styles and silhouettes. [ABSTRACT FROM AUTHOR]

Background: Lung cancer, also known as lung carcinoma, has a high mortality rate; however, an early prediction helps to reduce the risk. In the current literature, various approaches have been developed for the prediction of lung carcinoma (at an early stage), but these still have various issues, such as low accuracy, high noise, low contrast, poor recognition rates, and a high false-positive rate, etc. Thus, in this research effort, we have proposed an advanced algorithm and combined two different types of deep neural networks to make it easier to spot lung melanoma in the early phases. Methods: We have used WDSI (weakly supervised dense instance-level lung segmentation) for laborious pixel-level annotations. In addition, we suggested an SS-CL (deep continuous learning-based deep neural network) that can be applied to the labeled and unlabeled data to improve efficiency. This work intends to evaluate potential lightweight, low-memory deep neural net (DNN) designs for image processing. Results: Our experimental results show that, by combining WDSI and LSO segmentation, we can achieve super-sensitive, specific, and accurate early detection of lung cancer. For experiments, we used the lung nodule (LUNA16) dataset, which consists of the patients' 3D CT scan images. We confirmed that our proposed model is lightweight because it uses less memory. We have compared them with state-of-the-art models named PSNR and SSIM. The efficiency is 32.8% and 0.97, respectively. The proposed lightweight deep neural network (DNN) model archives a high accuracy of 98.2% and also removes noise more effectively. Conclusions: Our proposed approach has a lot of potential to help medical image analysis to help improve the accuracy of test results, and it may also prove helpful in saving patients' lives. [ABSTRACT FROM AUTHOR]

As a specialized map product, Wemaps must comply with relevant laws and regulations. Map audit plays a crucial role in ensuring map quality by preventing the production and dissemination of problem maps, as well as safeguarding national sovereignty, security, and interests. The user base for Wemaps is diverse, encompassing various types of maps, vast amounts of map data, and high expectations for timely dissemination. However, the current map audit process is inefficient and burdensome, failing to meet the specific needs of Wemaps audits. The key to solving this problem lies in the ability to automate and rapidly assess the audit requirements of Wemaps, approving those that require audit and promptly releasing those that do not. This study aims to establish an automated Wemaps audit assessment model using convolutional neural networks and transfer learning methods. By doing so, the burden of map audit can be reduced, and dissemination efficiency can be improved. The main contributions of this study are as follows: (1) Establishment of a dataset for assessing Wemaps audit requirements. (2) Utilization of VGG16 and ResNet50 neural network models for assessing Wemaps audit requirements; (3) Development of an optimal Wemaps audit assessment model through various experiments and training methods. (4) Analysis of factors influencing audit assessments based on measurement indicators and visualized results of the model. The experiments demonstrate that this method achieves high accuracy and can provide assessment services for public map audit requirements. Key policy highlights: A dataset for assessing Wemaps audit requirements is constructed Wemaps audit requirements are assessed using convolutional neural networks Visual variables Influencing the assessment of Wemaps audit requirements are analyzed [ABSTRACT FROM AUTHOR]

Automatic Speech Emotion Recognition (SER) plays a vital role in making human–computer interactions more natural and effective. A significant challenge in SER development is the limited availability of diverse emotional speech datasets, which hinders the application of advanced deep learning models. Transfer learning is a machine learning technique that helps address this issue by utilizing knowledge from pre-trained models to improve performance on a new task in a target domain, even with limited data. This study investigates the use of transfer learning from various pre-trained networks, including speaker embedding models such as d-vector, x-vector, and r-vector, and image classification models like AlexNet, GoogLeNet, SqueezeNet, ResNet-18, and ResNet-50. We also propose enhanced versions of the x-vector and r-vector models incorporating Multi-Head Attention Pooling and Angular Margin Softmax, alongside other architectural improvements. Additionally, reverberation from the Room Impulse Response datasets was added to the speech utterances to diversify and augment the available data. Notably, the enhanced r-vector model achieved classification accuracies of 74.05% Unweighted Accuracy (UA) and 73.68% Weighted Accuracy (WA) on the IEMOCAP dataset, and 80.25% UA and 79.81% WA on the CREMA-D dataset, outperforming the existing state-of-the-art methods. This study shows that using cross-domain transfer learning is beneficial for low-resource emotion recognition. The enhanced models developed in other domains (for non-emotional tasks) can further improve the accuracy of SER. [ABSTRACT FROM AUTHOR]

Compared to traditional technology, image classification technology possesses a superior capability for quantitative analysis of the target and background, and holds significant applications in the domains of ground target reconnaissance, marine environment monitoring, and emergency response to sudden natural disasters, among others. Currently, the enhancement of spatial spectral resolution heightens the difficulty and reduces the efficiency of classification, posing a substantial challenge to the aforementioned applications. Hence, the classification algorithm is required to take both computing power and classification accuracy into account. Research indicates that the deep kernel mapping network can accommodate both computing power and classification accuracy. By employing the kernel mapping function as the network node function of deep learning, it effectively enhances the classification accuracy under the condition of limited computing power. Therefore, to address the issue of network structure optimization of deep mapping networks and the insufficient application of line feature learning and expression in existing network structures, considering the adaptive optimization of network structures, deep quasiconformal kernel network learning (DQKNet) is proposed for image classification. Firstly, the structural parameters and learning parameters of the deep kernel mapping network are optimized. This approach can adaptively adjust the network structure based on the distribution characteristics of the data and enhance the performance of image classification. Secondly, the computational network node optimization method of quasiconformal kernel learning is applied to this network, further elevating the performance of the deep kernel learning mapping network in image classification. The experimental results demonstrate that the improvement in the deep kernel mapping network from the perspectives of accounting children, mapping network nodes, and network structure can effectively enhance the feature extraction and classification performance of the data. On the five public datasets, the average AA, OA, and KC values of our algorithm are 91.99, 91.25, and 85.99, respectively, outperforming the currently most-advanced algorithms. [ABSTRACT FROM AUTHOR]

Large variations across images make image classification a challenging task; limited training data further increases its difficulty. Genetic programming (GP) has been considerably applied to image classification. However, most GP methods tend to directly evolve a single classifier or depend on a predefined classification algorithm, which typically does not lead to ideal generalization performance when only a few training instances are available. Applying ensemble learning to classification often outperforms employing a single classifier. However, single-tree representation (each individual contains a single tree) is widely employed in GP. Training multiple diverse and accurate base learners/classifiers based on single-tree GP is challenging. Therefore, this article proposes a new ensemble construction method based on multi-tree GP (each individual contains multiple trees) for image classification. A single individual forms an ensemble, and its multiple trees constitute base learners. To find the best individual in which multiple trees are diverse and effectively cooperate, i.e., the nth tree can correct the errors of the previous n-1 trees, the new method assigns different weights to multiple trees using the idea of AdaBoost and performs classification via weighted majority voting. Furthermore, a new tree representation is developed to evolve diverse and accurate base learners that extract useful features and conduct classification simultaneously. The new approach achieves significantly better performance than almost all benchmark methods on eight datasets. Additional analyses highlight the effectiveness of the new ensembles and tree representation, demonstrating the potential for providing valuable interpretability in ensemble trees. [ABSTRACT FROM AUTHOR]

Nuclear-norm-based matrix regression (NMR) approaches utilize the nuclear norm for error term characterization, which strengthens the robustness of algorithm. However, NMR ignores the differences between samples from different classes, which leads to a poor feature representation. Moreover, NMR does not consider variations within different class, which affects the classification performance when classes are not homogeneous. To solve above problems, a local weighted representation-based matrix regression method (LWMR) is proposed. LWMR method solves two issues of current NMR methods that are based on nuclear norm. First, LWMR utilizes the prior distance information between test and training samples as weights, which improves the inter-class separation. Second, LWMR creates a new dictionary by averaging samples within different class and choosing the best representative sample for each class, which reduces the dictionary size and complexity. Experimental results on four widely used datasets demonstrate that LWMR method has faster calculation speed and better image performance than other regression models. [ABSTRACT FROM AUTHOR]

Quality assessment is a cornerstone of fruit production and distribution, particularly regarding storage conditions and duration. Citrus fruits, a staple in global consumption patterns, are the ultimate example. This study employs a nondestructive analytical technique, X-ray computed tomography (CT) scanning, to meticulously analyze a substantial sample of 300 citrus fruits, specifically satsuma, subjected to both ambient (20–22 °C, 50–60% humidity) and refrigeration conditions (6–8 °C, 65–75% humidity). The experiment was conducted through a methodologically rigorous approach, stratified dataset splitting, allocating 60% of the X-ray datasets for training, with 20% dedicated to validation and testing, respectively. The proposed research introduces a pioneering methodology termed features enhanced vision transformer (FEViT), meticulously designed to augment precision in citrus fruit classification and more precise freshness level prediction via X-ray image analysis. Our empirical findings unequivocally demonstrate the superior efficacy of FEViT models vis-a-vis conventional ViT counterparts across new X-ray citrus fruit datasets. Particularly noteworthy are the marked performance gains exhibited by FEViT-large variants, evidenced by notable increases in precision (5.08%), accuracy (5.47%), recall (4.55%), and F1 scores (5.28%) over original variants. This underscores the distinguishable enhanced discriminatory prowess of FEViT models in assessing citrus fruit quality in terms of freshness. Extensive validation through rigorous experimentation ratifies FEViT's supremacy over traditional deep learning architectures, affirming heightened accuracy (99.25%). The current study heralds the advent of FEViT architecture as a milestone in citrus fruit (satsuma) freshness prediction, promising augmented accuracy and robustness vis-a-vis extant methodologies. This research holds profound implications for the agricultural sector, especially in domains such as citrus and broader fruit classification, where nuanced image analysis is indispensable for quality attribute like freshness evaluation and informed decision-making. [ABSTRACT FROM AUTHOR]

The Explainable Artificial Intelligence (XAI) is a set of techniques and methods designed to make machine learning models and AI systems more transparent and interpretable. The goal of XAI is to enable humans to understand and trust the decisions made by AI systems, especially in critical or sensitive applications. In the context of generating explanations, XAI aims to provide human-understandable reasons for making a particular prediction or decision based on Artificial Intelligence (AI) model. XAI is an evolving field where the researchers have different ideas. Shapley values, which help explain the predictions, can be slow in calculation because they need to be figured out in many ways for each prediction. The current techniques for explaining AI are not fast and can be costly. This means they might not work well when trying to explain a lot of predictions at once. To overcome these issues, there developed a novel Convolution Bat Optimization based SHapley Additive exPlanations (CBO-SHAP) algorithm. Initially, a dataset contains a large number of images that are collected and pre-processed, then the data got divided into training, testing sets, and validation. Bat optimization is used for the classification and segmentation of the images which is performed in the pooling layer. To create the textual representation during testing, the trained images were applied to XAI with SHAP algorithm. Later, the image explanations were translated into the textual explanation that is readable by humans. This process enhances the whole operation with very less execution time of 78.3427sec. Comparative analyses against existing methods confirm the superior performance of our model, boasting high accuracy, f-measure, recall, and precision rates of approximately 99%, 99.28%, 99.22%, and 99.1% respectively. [ABSTRACT FROM AUTHOR]

The diagnosis is influenced by the classification of brain MRIs. Classifying and analyzing structures within images can be significantly enhanced by employing the Swin Transformer. The Swin Transformer is capable of capturing long-range relationships between pixels and creating layered image representations, which enhances its capacity to accurately evaluate brain structures. In addition, it has been demonstrated to be highly efficient in terms of computational power and memory usage, rendering it an optimal choice for high-resolution image processes. It is a tool that is adaptable in the field of computer vision, as it can be used for tasks such as object detection and image segmentation, demonstrating its versatility. The objective of this research is to identify tumors by classifying brain MRI image. In the event that tumors are identified, the research classifies them into three types: glioma, pituitary, and meningioma. The model utilizes 30,405 MRI images collected from two sources: Kaggle and the Al-Razi Medical Center in Iraq. The Swin Transformer's performance is enhanced by the integration of the feature selection mechanism with Gray Wolf Optimizer (GWO). The reduction in the number of features used during training, as a result of feature selection facilitated by GWO, leads to improved model efficiency and reduced data processing requirements. This simplified approach accelerates model performance by improving efficiency and memory usage. Additionally, optimal feature selection eliminates features that improve the models' accuracy and their capacity to differentiate between classes in MRI images. Incorporating the Fuzzy C Means (FCM) technique into feature selection may additionally improve the performance of GWO. The FCM assists in the grouping of related features to facilitate the selection of features by GWO. This method enhances the discrimination of non-useful features, resulting in reduced feature conflicts and improved model accuracy. Additionally, FCM assists in the identification of cluster centers that direct the GWO during optimization, thereby reducing the necessity for feature evaluation and improving processing efficiency. The optimization process is also expedited by integration. Improves the capacity of models to generalize in order to reduce the risk of overfitting and guarantee stability across datasets. This method enhances diagnostic accuracy by collecting features and analyzing structures in brain images, thereby providing a more profound understanding of the data. The proposed model achieved an accuracy of 99.490% with a loss rate of 0.0222 when trained on fuzzy optimization features. For training on fuzzy features derived from fuzzy wavelet MRI images, the model reached an accuracy of 99.572% and maintained a loss rate of 0.0206. [ABSTRACT FROM AUTHOR]

In this paper, we propose a novel imaging method to forecast the daily price data of West Texas Intermediate (WTI) crude oil futures. We use convolutional neural networks (CNNs) for future price trend prediction and obtain higher prediction accuracy than other benchmark forecasting methods. The results show that images can contain more nonlinear information, which is beneficial for energy price forecasting. Nonlinear factors also have a strong influence during drastic fluctuations in crude oil prices. In the robustness tests, we find that the image‐based CNN is the most stable approach and can be applied in various futures forecasting scenarios. In the prediction of low‐frequency models for high‐frequency data, the CNN method still retains considerable predictive power, indicating the possibility of transfer learning of our novel approach. By unleashing the power of the picture, we open up a whole new perspective for forecasting future energy trends. [ABSTRACT FROM AUTHOR]

In spatio-temporal plant monitoring, optical sensing (including hyperspectral imaging), is being deployed to, non-invasively, detect and diagnose plant responses to abiotic and biotic stressors. Early and accurate detection and diagnosis of stressors are key objectives. Level of radiometric repeatability of optical sensing data and ability to accurately detect and diagnose biotic stress are inversely correlated. Accordingly, it may be argued that one of the most significant frontiers and challenges regarding widespread adoption of optical sensing in plant research and crop production hinges on methods to maximize radiometric repeatability. In this study, we acquired hyperspectral optical sensing data at noon and midnight from soybean (Glycine max) and coleus wizard velvet red (Solenostemon scutellarioides) plants with/without experimentally infestation of two-spotted spider mites (Tetranychus urticae). We addressed three questions related to optimization of radiometric repeatability: (1) are reflectance-based plant responses affected by time of optical sensing? (2) if so, are plant responses to two-spotted spider mite infestations (biotic stressor) more pronounced at midnight versus at noon? (3) Is detection of biotic stress enhanced by spatial binning (smoothing) of hyperspectral imaging data? Results from this study provide insight into calculations of radiometric repeatability. Results strongly support claims that acquisition of optical sensing data to detect and characterize stress responses by plants to detect biotic stressors should be performed at night. Moreover, the combination of midnight imaging and spatial binning increased classification accuracies with 29% and 31% for soybean and coleus, respectively. Practical implications of these findings are discussed. Study results are relevant to virtually all applications of optical sensing to detect and diagnose abiotic and biotic stress responses by plants in both controlled environments and in outdoor crop production systems. [ABSTRACT FROM AUTHOR]

This study aims to explore an autoencoder-based method for generating brain MRI images of patients with Autism Spectrum Disorder (ASD) and non-ASD individuals, and to discriminate ASD based on the generated images. Initially, we introduce the research background of ASD and related work, as well as the application of deep learning in the field of medical imaging. Subsequently, we detail the architecture and training process of the proposed autoencoder model, and present the results of generating MRI images for ASD and non-ASD patients. Following this, we designed an ASD classifier based on the generated images and elucidated its structure and training methods. Finally, through analysis and discussion of experimental results, we validated the effectiveness of the proposed method and explored future research directions and potential clinical applications. This research offers new insights and methodologies for addressing challenges in ASD studies using deep learning technology, potentially contributing to the automated diagnosis and research of ASD. [ABSTRACT FROM AUTHOR]

The difficulty of selecting features efficiently in histopathology image analysis remains unresolved. Furthermore, the majority of current approaches have approached feature selection as a single objective issue. This research presents an enhanced multi-objective whale optimisation algorithm-based feature selection technique as a solution. To mine optimal feature sets, the suggested technique makes use of a unique variation known as the enhanced multi-objective whale optimisation algorithm. To verify the optimisation capability, the suggested variation has been evaluated on 10 common multi-objective CEC2009 benchmark functions. Furthermore, by comparing five classifiers in terms of accuracy, mean number of selected features, and calculation time, the effectiveness of the suggested strategy is verified against three other feature-selection techniques already in use. The experimental findings show that, when compared to the other approaches under consideration, the suggested method performed better on the assessed parameters. [ABSTRACT FROM AUTHOR]

Accurate information retrieval from multi-source and multi-resolution image data constitutes a foundation for knowledge discovery. Scene image classification in the remote sensing (RS) community using aerial very high resolution (VHR) images is one of the well-researched areas, which mostly utilise deep learning (DL)—based methods thanks to their remarkable classification performance. Nevertheless, existing DL-based methods still have a limited ability to capture precise spatial semantic information scattered toward the horizontal and vertical directions across such images at multiple scales and rotations. As such, we herein propose a novel approach, employing an innovative rotation invariant horizontal vertical pooled module (RIHVPM), to well-represent aerial VHR RS images for stable and improved classification performance. Notably, the proposed RIHVPM benefits from the multiple tensor rotations coupled with attention-enabled multiscale horizontal and vertical pooling operations for image representation. An experimental study on three benchmark datasets demonstrates competent and/or higher classification performance (AID: 96.44%, NWPU: 94.32% and UCM: 99.04%) and robustness/stability (minimum standard deviation of 0.001) of the proposed approach. [ABSTRACT FROM AUTHOR]

Background: The detection of lung nodules at their early stages may significantly enhance the survival rate and prevent progression to severe disability caused by advanced lung cancer, but it often requires manual and laborious efforts for radiologists, with limited success. To alleviate it, we propose a Multi-View Soft Attention-Based Convolutional Neural Network (MVSA-CNN) model for multi-class lung nodular classifications in three stages (benign, primary, and metastatic). Methods: Initially, patches from each nodule are extracted into three different views, each fed to our model to classify the malignancy. A dataset, namely the Lung Image Database Consortium Image Database Resource Initiative (LIDC-IDRI), is used for training and testing. The 10-fold cross-validation approach was used on the database to assess the model's performance. Results: The experimental results suggest that MVSA-CNN outperforms other competing methods with 97.10% accuracy, 96.31% sensitivity, and 97.45% specificity. Conclusions: We hope the highly predictive performance of MVSA-CNN in lung nodule classification from lung Computed Tomography (CT) scans may facilitate more reliable diagnosis, thereby improving outcomes for individuals with disabilities who may experience disparities in healthcare access and quality. [ABSTRACT FROM AUTHOR]

Currently, in the field of processing 3D point cloud data, two primary representation methods have emerged: point‐based methods and voxel‐based methods. However, the former suffer from significant computational costs and lack the ease of handling exhibited by voxel‐based methods. Conversely, the later often encounter challenges related to information loss resulting from downsampling operations, thereby impeding subsequent tasks. To address these limitations, this article introduces a novel density‐based representation method for voxel partitioning. Additionally, a corresponding network structure is devised to extract features from this specific density representation, thereby facilitating the successful completion of classification tasks. The experiments are implemented on ModelNet40 and MNIST demonstrate that the proposed 3D convolution can achieve the‐state‐of‐the‐art performance based on the voxels. [ABSTRACT FROM AUTHOR]

Target detection in autonomous driving tasks presents a complex and critical challenge due to the diversity of targets and the intricacy of the environment. To address this issue, this paper proposes an enhanced YOLOv8 model. Firstly, the original large target detection head is removed and replaced with a detection head tailored for small targets and high‐level semantic details. Secondly, an adaptive feature fusion method is proposed, where input feature maps are processed using dilated convolutions with different dilation rates, followed by adaptive feature fusion to generate adaptive weights. Finally, an improved attention mechanism is incorporated to enhance the model's focus on target regions. Additionally, the impact of Group Shuffle Convolution (GSConv) on the model's detection speed is investigated. Validated on two public datasets, the model achieves a mean Average Precision (mAP) of 53.7% and 53.5%. Although introducing GSConv results in a slight decrease in mAP, it significantly improves frames per second. These findings underscore the effectiveness of the proposed model in autonomous driving tasks. [ABSTRACT FROM AUTHOR]

As the demand for renewable energy sources increases, the vulnerability of solar panels to lightning strikes becomes a critical concern. This research explores the correlation between lightning-induced voltage fluctuations and the resultant damage intensity on solar panels. The study adopts a systematic approach, first investigating the correlation between lightning-induced voltage assessment using 30kV, 60kV and 90 kV impulse voltage with multi-stage Marx impulse generator and the damage intensity on monocrystalline and polycrystalline solar panels. A portable active infrared thermography equipment with tCam-Mini was employed to conduct this research with wireless streaming. Furthermore, the study integrates Convolutional Neural Network (CNN)-based image classification techniques to enhance the efficiency of damage assessment. The results highlight the potential of neural networks in improving the accuracy and speed of image classification for damaged and undamaged samples. The findings contribute valuable insights into enhancing the resilience of solar panel systems against lightning strikes, ultimately advancing the reliability and sustainability of solar energy infrastructure. A new CNN model was developed to classify the images obtained from thermography with 90.21% accuracy for greyscale and 85.31% accuracy on thermal images. [ABSTRACT FROM AUTHOR]

In machine learning classification problems, it is overly assumed that training and test data must be drawn from the same locale and essentially follow the same distribution for optimal performance. However, acquiring sufficient labeled data from the same locale is impractical and labeling is resource-intensive. Domain Adaptation (DA) offers a reliable solution, but manual feature and parameter selection for multi-dimensional image data are also resource-intensive and time-consuming. To address these issues collectively, we propose "PSO-based Unified Framework for Unsupervised Domain Adaptation in Image Classification" (PSO-UDAIC), an unsupervised domain adaptation framework, that minimizes domain discrepancy between labeled and unlabeled data by learning a subspace applying various objective functions, making them appear from the same distribution. The PSO technique is utilized to automate the selection of the optimal common subset of features and objective function parameters, reducing manual effort and time. Extensive experiments on benchmark image datasets (Office+Caltech (SURF, DeCAF, VGG6), USPS+MNIST, COIL20, and VLCS) with varied images and settings across diverse image classification tasks which achieved average accuracies of 58.18%, 93.71%, 91.64%, 100%, 81.76%, and 67.38%, respectively, outperforming the base method LDAPL by 12%, 2.12%, 2.55%, 0.28%, 2.76%, and 14.61% showing PSO-UDAIC's superior performance compared to primitive and domain adaptation methods. Statistical paired t-tests reaffirmed the significantly improved performance of PSO-UDAIC over the compared methods. [ABSTRACT FROM AUTHOR]

Knowledge distillation (KD) techniques aim to transfer knowledge from complex teacher neural networks to simpler student networks. In this study, we propose a novel knowledge distillation method called Multiloss Joint Gradient Control Knowledge Distillation (MJKD), which functions by effectively combining feature- and logit-based knowledge distillation methods with gradient control. The proposed knowledge distillation method discretely considers the gradients of the task loss (cross-entropy loss), feature distillation loss, and logit distillation loss. The experimental results suggest that logits may contain more information and should, consequently, be assigned greater weight during the gradient update process in this work. The empirical findings on the CIFAR-100 and Tiny-ImageNet datasets indicate that MJKD generally outperforms traditional knowledge distillation methods, significantly enhancing the generalization ability and classification accuracy of student networks. For instance, MJKD achieves a 63.53% accuracy on Tiny-ImageNet for the ResNet18 MobileNetV2 pair. Furthermore, we present visualizations and analyses to explore its potential working mechanisms. [ABSTRACT FROM AUTHOR]

Convolutional neural networks (CNNs) are widely utilized in image classification. Nevertheless, CNNs typically require substantial computational resources, posing challenges for deployment on resource-constrained edge devices and limiting the spread of AI-driven applications. While various pruning approaches have been proposed to mitigate this issue, they often overlook a critical fact that edge devices are typically tasked with handling only a subset of classes rather than the entire set. Moreover, the specific combinations of subcategories that each device must discern vary, highlighting the need for fine-grained task-specific adjustments. Unfortunately, these oversights result in pruned models that still contain unnecessary category redundancies, thereby impeding the potential for further model optimization and lightweight design. To bridge this gap, we propose a task-level customized pruning (TLCP) method via utilizing task-level information, i.e., class combination information relevant to edge devices. Specifically, TLCP first introduces channel control gates to assess the importance of each convolutional channel for individual classes. These class-level control gates are then aggregated through linear combinations, resulting in a pruned model customized to the specific tasks of edge devices. Experiments on various customized tasks demonstrate that TLCP can significantly reduce the number of parameters, by up to 33.9% on CIFAR-10 and 14.0% on CIFAR-100, compared to other baseline methods, while maintaining almost the same inference accuracy. [ABSTRACT FROM AUTHOR]

Early detection of skin cancer significantly improves the five-year survival rate for patients. However, due to the subtle nature of early malignant tumors in the skin, their symptoms are not apparent, and specialized doctors need to perform multiple biopsies and extract lesion tissues to diagnose the type of lesion. Traditional artificial intelligence methods have low accuracy in identifying skin lesion images primarily because they are difficult to simultaneously focus on spatial details and shallow semantic features. To effectively represent spatial positions and shallow feature information, and avoid the model being overly concerned with detailed information, which can lead to misclassification of easily distinguishable images, a progressive aggregation attention network based on ConvNeXt is proposed. This method utilizes a hierarchical ConvNeXt encoder to extract deep and shallow features of lesion regions layer by layer. By employing parallel spatial attention, it effectively integrates spatial position information with deep or shallow semantic features, aggregating multi-scale contextual information. Meanwhile, a progressive feature aggregation module is designed to effectively integrate deep and shallow features and aggregate them by dynamically adjusting weights, closely aligning with the process of rough observation and meticulous examination in skin image classification by specialized doctors. Experimental tests on the ISIC2018 and ISIC2019 datasets show that the accuracy, precision, recall, and F1-Score of this method are 95.27%, 93.76%, 92.83%, and 93.18%, respectively, for ISIC2018, and 92.63%, 91.06%, 87.05%, and 88.81%, respectively, for ISIC2019. Compared to ConvNeXt, the accuracy is improved by 2.13 and 3.29 percentage points respectively, demonstrating its ability to effectively extract detailed and rough features, providing new evidence for the diagnosis of skin image through dermatoscopy. [ABSTRACT FROM AUTHOR]

Diabetes Mellitus (DM) is a chronic condition that affects the blood glucose metabolism of various organs and tissues throughout the body. It can result in microvascular disorders such as coronary heart disease and cerebral hemorrhage. One significant complication is retinopathy, which, in severe cases, can lead to blindness. Early screening and detection are crucial as the disease process is irreversible. In this study, we developed a model for early screening of Diabetes Retinopathy (DR) using color fundus photography images. Our approach involved employing CLAHE, grayscale image transformation methods, and transfer learning to improve diagnostic efficiency when working with limited data. The APTOS 2019 dataset, consisting of3662 retinal images, was used in this research. Four different preprocessing methods were applied to the retinal images, including removing the black edge, resizing, and normalization (Method I), adding contrast constrained adaptive histogram equalization (CLAHE) to Method I (Method II), adding grayscale transformation to Method I (Method III), and adding CLAHE and grayscale transformations to Method I (Method IV). Data augmentation techniques such as random brightness and contrast transformations, flipping, image cropping, and mix-up algorithms were utilized for data enhancement. The ResNet50 and InceptionV3 models based on convolutional neural networks were employed to build the model for learning retinal images under three scenarios: (1) learning from scratch, (2) transfer learning with fixed weights and training only the fully connected layer, and (3) transfer learning with loaded weights, followed by fine-tuning of the entire network based on the input data. The classification performance of the models was evaluated using metrics such as AUC, accuracy, F1 score, precision, and recall. For the ResNet50 model, the accuracy rates for learning from scratch, fixed weight, and fine-tuning weight were 75.41%, 54.64%, and 81.97%, respectively. When using the InceptionV3 model, the accuracy rates were 76.50%, 10.38%, and 83.61%, respectively. Fine-tuning was conducted on data II, III, and IV using the InceptionV3 model, resulting in accuracies of 81.42%, 80.87%, and 83.61%, respectively. Comparisons between models using the same data and training methods revealed that models employing the InceptionV3 structure achieved higher accuracy than those using ResNet50 (83.61% vs. 81.97%). The results indicate that the InceptionV3-based CNN, coupled with transfer learning and appropriate data pre-processing methods, exhibited superior performance in accurately detecting diabetic retinopathy, as measured by accuracy, AUC, F1 score, and other evaluation metrics. This research holds significant value in enabling efficient early diagnosis of DR lesions and conducting an intelligent and efficient graded diagnosis of the DR progression, thereby providing the groundwork for timely intervention. [ABSTRACT FROM AUTHOR]

Image classification is a crucial task in machine learning with widespread practical applications. The existing classical framework for image classification typically utilizes a global pooling operation at the end of the network to reduce computational complexity and mitigate overfitting. However, this operation often results in a significant loss of information, which can affect the performance of classification models. To overcome this limitation, we introduce a novel image classification framework that leverages variational quantum algorithms (VQAs) hybrid approaches combining quantum and classical computing paradigms within quantum machine learning. The major advantage of our framework is the elimination of the need for the global pooling operation at the end of the network. In this way, our approach preserves more discriminative features and fine-grained details in the images, which enhances classification performance. Additionally, employing VQAs enables our framework to have fewer parameters than the classical framework, even in the absence of global pooling, which makes it more advantageous in preventing overfitting. We apply our method to different state-of-the-art image classification models and demonstrate the superiority of the proposed quantum architecture over its classical counterpart through a series of state vector simulation experiments on public datasets. Our experiments show that the proposed quantum framework achieves up to a 9.21% increase in accuracy and up to a 15.79% improvement in F1 score, compared to the classical framework. Additionally, we explore the impact of shot noise on our method through shot-based simulation and find that increasing the number of measurements does not always lead to better results. Selecting an appropriate number of measurements can yield optimal results, even surpassing those obtained from state vector simulation. [ABSTRACT FROM AUTHOR]

The application of convolutional neural network has been gradually introduced in the field of X-ray bone images. At present, there is less research on methods to automatically detect abnormal parts of the skeleton. The method improves on the Densenet network by adjusting the network structure, adding a convolutional block attention module (CBAM) attention mechanism to the original Densenlayer, fusing spatial attention and channel attention to suppress unnecessary features and strengthen the network's capability to extract image features, and optimizing the Transition block by fusing two pooling strategies, average and maximum, to increase the model's anti-interference capability. The data collected after the experiment show that the improved Densenet has increased accuracy in the detection of skeletal abnormalities at all sites compared with the traditional network. The average accuracy is 1–5% higher than the benchmark method, the highest elbow accuracy reached 77.62%, the lowest hand accuracy also reached 64.28% and the area under the receiver operating characteristic curve is increased by 3–7%, the highest reached 80.83%. [ABSTRACT FROM AUTHOR]

Using a deep learning method to classify a large amount of labelled remote sensing scene data produces good performance. However, it is challenging for deep learning based methods to generalise to classification tasks with limited data. Few‐shot learning allows neural networks to classify unseen categories when confronted with a handful of labelled data. Currently, episodic tasks based on meta‐learning can effectively complete few‐shot classification, and training an encoder that can conduct representation learning has become an important component of few‐shot learning. An end‐to‐end few‐shot remote sensing scene classification model based on ProtoNet and self‐supervised learning is proposed. The authors design the Pre‐prototype for a more discrete feature space and better integration with self‐supervised learning, and also propose the ProtoMixer for higher quality prototypes with a global receptive field. The authors' method outperforms the existing state‐of‐the‐art self‐supervised based methods on three widely used benchmark datasets: UC‐Merced, NWPU‐RESISC45, and AID. Compare with previous state‐of‐the‐art performance. For the one‐shot setting, this method improves by 1.21%, 2.36%, and 0.84% in AID, UC‐Merced, and NWPU‐RESISC45, respectively. For the five‐shot setting, this method surpasses by 0.85%, 2.79%, and 0.74% in the AID, UC‐Merced, and NWPU‐RESISC45, respectively. [ABSTRACT FROM AUTHOR]

Siamese networks excel at comparing two images, serving as an effective class verification technique for a single‐per‐class reference image. However, when multiple reference images are present, Siamese verification necessitates multiple comparisons and aggregation, often unpractical at inference. The Centre‐Loss approach, proposed in this research, solves a class verification task more efficiently, using a single forward‐pass during inference, than sample‐to‐sample approaches. Optimising a Centre‐Loss function learns class centres and minimises intra‐class distances in latent space. The authors compared verification accuracy using Centre‐Loss against aggregated Siamese when other hyperparameters (such as neural network backbone and distance type) are the same. Experiments were performed to contrast the ubiquitous Euclidean against other distance types to discover the optimum Centre‐Loss layer, its size, and Centre‐Loss weight. In optimal architecture, the Centre‐Loss layer is connected to the penultimate layer, calculates Euclidean distance, and its size depends on distance type. The Centre‐Loss method was validated on the Self‐Checkout products and Fruits 360 image datasets. Centre‐Loss comparable accuracy and lesser complexity make it a preferred approach over sample‐to‐sample for the class verification task, when the number of reference image per class is high and inference speed is a factor, such as in self‐checkouts. [ABSTRACT FROM AUTHOR]