Sentiment analysis, a crucial task in discerning emotional tones within the text, plays a pivotal role in understandingpublic opinion and user sentiment across diverse languages.While numerous scholars conduct sentime...Sentiment analysis, a crucial task in discerning emotional tones within the text, plays a pivotal role in understandingpublic opinion and user sentiment across diverse languages.While numerous scholars conduct sentiment analysisin widely spoken languages such as English, Chinese, Arabic, Roman Arabic, and more, we come to grapplingwith resource-poor languages like Urdu literature which becomes a challenge. Urdu is a uniquely crafted language,characterized by a script that amalgamates elements from diverse languages, including Arabic, Parsi, Pashtu,Turkish, Punjabi, Saraiki, and more. As Urdu literature, characterized by distinct character sets and linguisticfeatures, presents an additional hurdle due to the lack of accessible datasets, rendering sentiment analysis aformidable undertaking. The limited availability of resources has fueled increased interest among researchers,prompting a deeper exploration into Urdu sentiment analysis. This research is dedicated to Urdu languagesentiment analysis, employing sophisticated deep learning models on an extensive dataset categorized into fivelabels: Positive, Negative, Neutral, Mixed, and Ambiguous. The primary objective is to discern sentiments andemotions within the Urdu language, despite the absence of well-curated datasets. To tackle this challenge, theinitial step involves the creation of a comprehensive Urdu dataset by aggregating data from various sources such asnewspapers, articles, and socialmedia comments. Subsequent to this data collection, a thorough process of cleaningand preprocessing is implemented to ensure the quality of the data. The study leverages two well-known deeplearningmodels, namely Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN), for bothtraining and evaluating sentiment analysis performance. Additionally, the study explores hyperparameter tuning tooptimize the models’ efficacy. Evaluation metrics such as precision, recall, and the F1-score are employed to assessthe effectiveness of the models. The research findings reveal that RNN surpasses CNN in Urdu sentiment analysis,gaining a significantly higher accuracy rate of 91%. This result accentuates the exceptional performance of RNN,solidifying its status as a compelling option for conducting sentiment analysis tasks in the Urdu language.展开更多
Underground engineering projects such as deep tunnel excavation often encounter rockburst disasters accompanied by numerous microseismic events.Rapid interpretation of microseismic signals is crucial for the timely id...Underground engineering projects such as deep tunnel excavation often encounter rockburst disasters accompanied by numerous microseismic events.Rapid interpretation of microseismic signals is crucial for the timely identification of rockbursts.However,conventional processing encompasses multi-step workflows,including classification,denoising,picking,locating,and computational analysis,coupled with manual intervention,which collectively compromise the reliability of early warnings.To address these challenges,this study innovatively proposes the“microseismic stethoscope"-a multi-task machine learning and deep learning model designed for the automated processing of massive microseismic signals.This model efficiently extracts three key parameters that are necessary for recognizing rockburst disasters:rupture location,microseismic energy,and moment magnitude.Specifically,the model extracts raw waveform features from three dedicated sub-networks:a classifier for source zone classification,and two regressors for microseismic energy and moment magnitude estimation.This model demonstrates superior efficiency compared to traditional processing and semi-automated processing,reducing per-event processing time from 0.71 s to 0.49 s to merely 0.036 s.It concurrently achieves 98%accuracy in source zone classification,with microseismic energy and moment magnitude estimation errors of 0.13 and 0.05,respectively.This model has been well applied and validated in the Daxiagu Tunnel case in Sichuan,China.The application results indicate that the model is as accurate as traditional methods in determining source parameters,and thus can be used to identify potential geomechanical processes of rockburst disasters.By enhancing the signal processing reliability of microseismic events,the proposed model in this study presents a significant advancement in the identification of rockburst disasters.展开更多
The rapid growth of biomedical data,particularly multi-omics data including genomes,transcriptomics,proteomics,metabolomics,and epigenomics,medical research and clinical decision-making confront both new opportunities...The rapid growth of biomedical data,particularly multi-omics data including genomes,transcriptomics,proteomics,metabolomics,and epigenomics,medical research and clinical decision-making confront both new opportunities and obstacles.The huge and diversified nature of these datasets cannot always be managed using traditional data analysis methods.As a consequence,deep learning has emerged as a strong tool for analysing numerous omics data due to its ability to handle complex and non-linear relationships.This paper explores the fundamental concepts of deep learning and how they are used in multi-omics medical data mining.We demonstrate how autoencoders,variational autoencoders,multimodal models,attention mechanisms,transformers,and graph neural networks enable pattern analysis and recognition across all omics data.Deep learning has been found to be effective in illness classification,biomarker identification,gene network learning,and therapeutic efficacy prediction.We also consider critical problems like as data quality,model explainability,whether findings can be repeated,and computational power requirements.We now consider future elements of combining omics with clinical and imaging data,explainable AI,federated learning,and real-time diagnostics.Overall,this study emphasises the need of collaborating across disciplines to advance deep learning-based multi-omics research for precision medicine and comprehending complicated disorders.展开更多
Human Activity Recognition(HAR)is a novel area for computer vision.It has a great impact on healthcare,smart environments,and surveillance while is able to automatically detect human behavior.It plays a vital role in ...Human Activity Recognition(HAR)is a novel area for computer vision.It has a great impact on healthcare,smart environments,and surveillance while is able to automatically detect human behavior.It plays a vital role in many applications,such as smart home,healthcare,human computer interaction,sports analysis,and especially,intelligent surveillance.In this paper,we propose a robust and efficient HAR system by leveraging deep learning paradigms,including pre-trained models,CNN architectures,and their average-weighted fusion.However,due to the diversity of human actions and various environmental influences,as well as a lack of data and resources,achieving high recognition accuracy remain elusive.In this work,a weighted average ensemble technique is employed to fuse three deep learning models:EfficientNet,ResNet50,and a custom CNN.The results of this study indicate that using a weighted average ensemble strategy for developing more effective HAR models may be a promising idea for detection and classification of human activities.Experiments by using the benchmark dataset proved that the proposed weighted ensemble approach outperformed existing approaches in terms of accuracy and other key performance measures.The combined average-weighted ensemble of pre-trained and CNN models obtained an accuracy of 98%,compared to 97%,96%,and 95%for the customized CNN,EfficientNet,and ResNet50 models,respectively.展开更多
Recent advances in deep learning have significantly improved flood detection and segmentation from aerial and satellite imagery.However,conventional convolutional neural networks(CNNs)often struggle in complex flood s...Recent advances in deep learning have significantly improved flood detection and segmentation from aerial and satellite imagery.However,conventional convolutional neural networks(CNNs)often struggle in complex flood scenarios involving reflections,occlusions,or indistinct boundaries due to limited contextual modeling.To address these challenges,we propose a hybrid flood segmentation framework that integrates a Vision Transformer(ViT)encoder with a U-Net decoder,enhanced by a novel Flood-Aware Refinement Block(FARB).The FARB module improves boundary delineation and suppresses noise by combining residual smoothing with spatial-channel attention mechanisms.We evaluate our model on a UAV-acquired flood imagery dataset,demonstrating that the proposed ViTUNet+FARB architecture outperforms existing CNN and Transformer-based models in terms of accuracy and mean Intersection over Union(mIoU).Detailed ablation studies further validate the contribution of each component,confirming that the FARB design significantly enhances segmentation quality.To its better performance and computational efficiency,the proposed framework is well-suited for flood monitoring and disaster response applications,particularly in resource-constrained environments.展开更多
The estimation of the Number of Sources(NoS)is a significant challenge in signal processing,particularly due to the impact of colored noise on the performance of NoS estimation.This paper proposes a Multidimensional F...The estimation of the Number of Sources(NoS)is a significant challenge in signal processing,particularly due to the impact of colored noise on the performance of NoS estimation.This paper proposes a Multidimensional Feature Network(MFNet)which is designed for NoS estimation by extracting features of the sampled received signals and Sampled Covariance Matrix(SCM).The MFNet treats the raw signal and the SCM as two different types of data,and is able to achieve NoS estimation under colored noise and imperfect array.MFNet employs the Gated Recurrent Unit(GRU)to capture sequential information from the original signal data and to construct the Pseudo Covariance Matrix(PCM).Subsequently,various dimensional features,including eigenvalues and the Gerschgorin disk radius,are extracted from both the PCM and SCM,which are then jointly input into the subsequent network.An overall accuracy of 82%can be achieved after network training.The ablation experimental results demonstrate the effectiveness of multiple inputs.And simulation results demonstrate that the proposed MFNet achieves higher estimation accuracy compared to existing algorithms and exhibits greater robustness against colored noise.展开更多
This study focuses on developing a deep learning model capable of recognizing vehicle brands and models,integrated with a law enforcement intelligence platform to overcome the limitations of existing license plate rec...This study focuses on developing a deep learning model capable of recognizing vehicle brands and models,integrated with a law enforcement intelligence platform to overcome the limitations of existing license plate recognition techniques—particularly in handling counterfeit,obscured,or absent plates.The research first entailed collecting,annotating,and classifying images of various vehiclemodels,leveraging image processing and feature extraction methodologies to train themodel on Microsoft Custom Vision.Experimental results indicate that,formost brands and models,the system achieves stable and relatively high performance in Precision,Recall,and Average Precision(AP).Furthermore,simulated tests involving illicit vehicles reveal that,even in cases of reassigned,concealed,or missing license plates,the model can rely on exterior body features to effectively identify vehicles,reducing dependence on plate-specific data.In practical law enforcement scenarios,these findings can accelerate investigations of stolen or forged plates and enhance overall accuracy.In conclusion,continued collection of vehicle images across broadermodel types,production years,and modification levels—along with refined annotation processes and parameter adjustment strategies—will further strengthen themethod’s applicability within law enforcement intelligence platforms,facilitating more precise and comprehensive vehicle recognition and control in real-world operations.展开更多
Near real-time maize phenology monitoring is crucial for field management,cropping system adjustments,and yield estimation.Most phenological monitoring methods are post-seasonal and heavily rely on high-frequency time...Near real-time maize phenology monitoring is crucial for field management,cropping system adjustments,and yield estimation.Most phenological monitoring methods are post-seasonal and heavily rely on high-frequency time-series data.These methods are not applicable on the unmanned aerial vehicle(UAV)platform due to the high cost of acquiring time-series UAV images and the shortage of UAV-based phenological monitoring methods.To address these challenges,we employed the Synthetic Minority Oversampling Technique(SMOTE)for sample augmentation,aiming to resolve the small sample modelling problem.Moreover,we utilized enhanced"separation"and"compactness"feature selection methods to identify input features from multiple data sources.In this process,we incorporated dynamic multi-source data fusion strategies,involving Vegetation index(VI),Color index(CI),and Texture features(TF).A two-stage neural network that combines Convolutional Neural Network(CNN)and Long Short-Term Memory Network(LSTM)is proposed to identify maize phenological stages(including sowing,seedling,jointing,trumpet,tasseling,maturity,and harvesting)on UAV platforms.The results indicate that the dataset generated by SMOTE closely resembles the measured dataset.Among dynamic data fusion strategies,the VI-TF combination proves to be most effective,with CI-TF and VI-CI combinations following behind.Notably,as more data sources are integrated,the model's demand for input features experiences a significant decline.In particular,the CNN-LSTM model,based on the fusion of three data sources,exhibited remarkable reliability when validating the three datasets.For Dataset 1(Beijing Xiaotangshan,2023:Data from 12 UAV Flight Missions),the model achieved an overall accuracy(OA)of 86.53%.Additionally,its precision(Pre),recall(Rec),F1 score(F1),false acceptance rate(FAR),and false rejection rate(FRR)were 0.89,0.89,0.87,0.11,and 0.11,respectively.The model also showed strong generalizability in Dataset 2(Beijing Xiaotangshan,2023:Data from 6 UAV Flight Missions)and Dataset 3(Beijing Xiaotangshan,2022:Data from 4 UAV Flight Missions),with OAs of 89.4%and 85%,respectively.Meanwhile,the model has a low demand for input featu res,requiring only 54.55%(99 of all featu res).The findings of this study not only offer novel insights into near real-time crop phenology monitoring,but also provide technical support for agricultural field management and cropping system adaptation.展开更多
The deep learning algorithm,which has been increasingly applied in the field of petroleum geophysical prospecting,has achieved good results in improving efficiency and accuracy based on test applications.To play a gre...The deep learning algorithm,which has been increasingly applied in the field of petroleum geophysical prospecting,has achieved good results in improving efficiency and accuracy based on test applications.To play a greater role in actual production,these algorithm modules must be integrated into software systems and used more often in actual production projects.Deep learning frameworks,such as TensorFlow and PyTorch,basically take Python as the core architecture,while the application program mainly uses Java,C#,and other programming languages.During integration,the seismic data read by the Java and C#data interfaces must be transferred to the Python main program module.The data exchange methods between Java,C#,and Python include shared memory,shared directory,and so on.However,these methods have the disadvantages of low transmission efficiency and unsuitability for asynchronous networks.Considering the large volume of seismic data and the need for network support for deep learning,this paper proposes a method of transmitting seismic data based on Socket.By maximizing Socket’s cross-network and efficient longdistance transmission,this approach solves the problem of inefficient transmission of underlying data while integrating the deep learning algorithm module into a software system.Furthermore,the actual production application shows that this method effectively solves the shortage of data transmission in shared memory,shared directory,and other modes while simultaneously improving the transmission efficiency of massive seismic data across modules at the bottom of the software.展开更多
The accurate prediction of battery pack capacity in electric vehicles(EVs)is crucial for ensuring safety and optimizing performance.Despite extensive research on predicting cell capacity using laboratory data,predicti...The accurate prediction of battery pack capacity in electric vehicles(EVs)is crucial for ensuring safety and optimizing performance.Despite extensive research on predicting cell capacity using laboratory data,predicting the capacity of onboard battery packs from field data remains challenging due to complex operating conditions and irregular EV usage in real-world settings.Most existing methods rely on extracting health feature parameters from raw data for capacity prediction of onboard battery packs,however,selecting specific parameters often results in a loss of critical information,which reduces prediction accuracy.To this end,this paper introduces a novel framework combining deep learning and data compression techniques to accurately predict battery pack capacity onboard.The proposed data compression method converts monthly EV charging data into feature maps,which preserve essential data characteristics while reducing the volume of raw data.To address missing capacity labels in field data,a capacity labeling method is proposed,which calculates monthly battery capacity by transforming the ampere-hour integration formula and applying linear regression.Subsequently,a deep learning model is proposed to build a capacity prediction model,using feature maps from historical months to predict the battery capacity of future months,thus facilitating accurate forecasts.The proposed framework,evaluated using field data from 20 EVs,achieves a mean absolute error of 0.79 Ah,a mean absolute percentage error of 0.65%,and a root mean square error of 1.02 Ah,highlighting its potential for real-world EV applications.展开更多
Many bioinformatics applications require determining the class of a newly sequenced Deoxyribonucleic acid(DNA)sequence,making DNA sequence classification an integral step in performing bioinformatics analysis,where la...Many bioinformatics applications require determining the class of a newly sequenced Deoxyribonucleic acid(DNA)sequence,making DNA sequence classification an integral step in performing bioinformatics analysis,where large biomedical datasets are transformed into valuable knowledge.Existing methods rely on a feature extraction step and suffer from high computational time requirements.In contrast,newer approaches leveraging deep learning have shown significant promise in enhancing accuracy and efficiency.In this paper,we investigate the performance of various deep learning architectures:Convolutional Neural Network(CNN),CNN-Long Short-Term Memory(CNNLSTM),CNN-Bidirectional Long Short-Term Memory(CNN-BiLSTM),Residual Network(ResNet),and InceptionV3 for DNA sequence classification.Various numerical and visual data representation techniques are utilized to represent the input datasets,including:label encoding,k-mer sentence encoding,k-mer one-hot vector,Frequency Chaos Game Representation(FCGR)and 5-Color Map(ColorSquare).Three datasets are used for the training of the models including H3,H4 and DNA Sequence Dataset(Yeast,Human,Arabidopsis Thaliana).Experiments are performed to determine which combination of DNA representation and deep learning architecture yields improved performance for the classification task.Our results indicate that using a hybrid CNN-LSTM neural network trained on DNA sequences represented as one-hot encoded k-mer sequences yields the best performance,achieving an accuracy of 92.1%.展开更多
Bone tumors(BTs)-including osteosarcoma,Ewing sarcoma,and chondrosarcoma-are rare but biologically complex malignancies characterized by pronounced heterogeneity in anatomical location,histological subtype,and molecul...Bone tumors(BTs)-including osteosarcoma,Ewing sarcoma,and chondrosarcoma-are rare but biologically complex malignancies characterized by pronounced heterogeneity in anatomical location,histological subtype,and molecular alterations.Recent advances in artificial intelligence(AI),particularly deep learning,have enabled the integration of diverse clinical data modalities to support diagnosis,treatment planning,and prognostication in bone oncology.This review provides a comprehensive synthesis of AI-driven multimodal fusion strategies that incorporate radiological imaging,digital pathology,multi-omics profiling,and electronic health records.We conducted a structured review of peer-reviewed literature published between 2015 and early 2025,focusing on the development,validation,and clinical applicability of AI models for BT diagnosis,subtyping,treatment response prediction,and recurrence monitoring.Although multimodal models have demonstrated advantages over unimodal approaches,especially in handling missing data and improving generalizability,most remain constrained by single-center study designs,small sample sizes,and limited prospective or external validation.Persistent technical and translational challenges include semantic misalignment across modalities,incomplete datasets,limited model interpretability,and regulatory and infrastructural barriers to clinical integration.To address these limitations,we highlight emerging directions such as contrastive representation learning,generative data augmentation,transformer-based fusion architectures,and privacy-preserving federated learning.We also discuss the evolving role of foundation models and workflow-integrated AI agents in enhancing scalability and clinical usability.In summary,multimodal AI represents a promising paradigm for advancing precision care in BTs.Realizing its full clinical potential will require methodologically rigorous,biologically informed,and system-level approaches that bridge algorithmic innovation with real-world healthcare delivery.展开更多
Efficient and accurate prediction of ocean surface latent heat fluxes is essential for understanding and modeling climate dynamics.Conventional estimation methods have low resolution and lack accuracy.The transformer ...Efficient and accurate prediction of ocean surface latent heat fluxes is essential for understanding and modeling climate dynamics.Conventional estimation methods have low resolution and lack accuracy.The transformer model,with its self-attention mechanism,effectively captures long-range dependencies,leading to a degradation of accuracy over time.Due to the non-linearity and uncertainty of physical processes,the transformer model encounters the problem of error accumulation,leading to a degradation of accuracy over time.To solve this problem,we combine the Data Assimilation(DA)technique with the transformer model and continuously modify the model state to make it closer to the actual observations.In this paper,we propose a deep learning model called TransNetDA,which integrates transformer,convolutional neural network and DA methods.By combining data-driven and DA methods for spatiotemporal prediction,TransNetDA effectively extracts multi-scale spatial features and significantly improves prediction accuracy.The experimental results indicate that the TransNetDA method surpasses traditional techniques in terms of root mean square error and R2 metrics,showcasing its superior performance in predicting latent heat fluxes at the ocean surface.展开更多
Deep learning now underpins many state-of-the-art systems for biomedical image and signal processing,enabling automated lesion detection,physiological monitoring,and therapy planning with accuracy that rivals expert p...Deep learning now underpins many state-of-the-art systems for biomedical image and signal processing,enabling automated lesion detection,physiological monitoring,and therapy planning with accuracy that rivals expert performance.This survey reviews the principal model families as convolutional,recurrent,generative,reinforcement,autoencoder,and transfer-learning approaches as emphasising how their architectural choices map to tasks such as segmentation,classification,reconstruction,and anomaly detection.A dedicated treatment of multimodal fusion networks shows how imaging features can be integrated with genomic profiles and clinical records to yield more robust,context-aware predictions.To support clinical adoption,we outline post-hoc explainability techniques(Grad-CAM,SHAP,LIME)and describe emerging intrinsically interpretable designs that expose decision logic to end users.Regulatory guidance from the U.S.FDA,the European Medicines Agency,and the EU AI Act is summarised,linking transparency and lifecycle-monitoring requirements to concrete development practices.Remaining challenges as data imbalance,computational cost,privacy constraints,and cross-domain generalization are discussed alongside promising solutions such as federated learning,uncertainty quantification,and lightweight 3-D architectures.The article therefore offers researchers,clinicians,and policymakers a concise,practice-oriented roadmap for deploying trustworthy deep-learning systems in healthcare.展开更多
Crop phenotyping plays a critical role in precision agriculture by enabling the accurate assessment of plant traits,supporting improved crop management,breeding programs,and yield optimization.However,cowpea leaves pr...Crop phenotyping plays a critical role in precision agriculture by enabling the accurate assessment of plant traits,supporting improved crop management,breeding programs,and yield optimization.However,cowpea leaves present unique challenges for automated phenotyping due to their diverse shapes,complex vein structures,and variations caused by environmental conditions.This research presents a deep learning-based approach for the classification of cowpea leaf images to support crop phenotyping tasks.Given the limited availability of annotated datasets,data augmentation techniques were employed to artificially expand the original small dataset while preserving essential leaf characteristics.Various image processing methods were applied to enrich the dataset,ensuring better feature representation without significant information loss.A deep neural network,specifically the MobileNet architecture,was utilized for its efficiency in capturing multi-scale features and handling image data with limited computational resources.The performance of the model trained on the augmented dataset was evaluated,achieving an accuracy of 94.12%on the cowpea leaf classification task.These results demonstrate the effectiveness of data augmentation in enhancing model generalization and learning capabilities.展开更多
Lung cancer continues to be a leading cause of cancer-related deaths worldwide,emphasizing the critical need for improved diagnostic techniques.Early detection of lung tumors significantly increases the chances of suc...Lung cancer continues to be a leading cause of cancer-related deaths worldwide,emphasizing the critical need for improved diagnostic techniques.Early detection of lung tumors significantly increases the chances of successful treatment and survival.However,current diagnostic methods often fail to detect tumors at an early stage or to accurately pinpoint their location within the lung tissue.Single-model deep learning technologies for lung cancer detection,while beneficial,cannot capture the full range of features present in medical imaging data,leading to incomplete or inaccurate detection.Furthermore,it may not be robust enough to handle the wide variability in medical images due to different imaging conditions,patient anatomy,and tumor characteristics.To overcome these disadvantages,dual-model or multi-model approaches can be employed.This research focuses on enhancing the detection of lung cancer by utilizing a combination of two learning models:a Convolutional Neural Network(CNN)for categorization and the You Only Look Once(YOLOv8)architecture for real-time identification and pinpointing of tumors.CNNs automatically learn to extract hierarchical features from raw image data,capturing patterns such as edges,textures,and complex structures that are crucial for identifying lung cancer.YOLOv8 incorporates multiscale feature extraction,enabling the detection of tumors of varying sizes and scales within a single image.This is particularly beneficial for identifying small or irregularly shaped tumors that may be challenging to detect.Furthermore,through the utilization of cutting-edge data augmentation methods,such as Deep Convolutional Generative Adversarial Networks(DCGAN),the suggested approach can handle the issue of limited data and boost the models’ability to learn from diverse and comprehensive datasets.The combined method not only improved accuracy and localization but also ensured efficient real-time processing,which is crucial for practical clinical applications.The CNN achieved an accuracy of 97.67%in classifying lung tissues into healthy and cancerous categories.The YOLOv8 model achieved an Intersection over Union(IoU)score of 0.85 for tumor localization,reflecting high precision in detecting and marking tumor boundaries within the images.Finally,the incorporation of synthetic images generated by DCGAN led to a 10%improvement in both the CNN classification accuracy and YOLOv8 detection performance.展开更多
To increase the storage capacity in holographic data storage(HDS),the information to be stored is encoded into a complex amplitude.Fast and accurate retrieval of amplitude and phase from the reconstructed beam is nece...To increase the storage capacity in holographic data storage(HDS),the information to be stored is encoded into a complex amplitude.Fast and accurate retrieval of amplitude and phase from the reconstructed beam is necessary during data readout in HDS.In this study,we proposed a complex amplitude demodulation method based on deep learning from a single-shot diffraction intensity image and verified it by a non-interferometric lensless experiment demodulating four-level amplitude and four-level phase.By analyzing the correlation between the diffraction intensity features and the amplitude and phase encoding data pages,the inverse problem was decomposed into two backward operators denoted by two convolutional neural networks(CNNs)to demodulate amplitude and phase respectively.The experimental system is simple,stable,and robust,and it only needs a single diffraction image to realize the direct demodulation of both amplitude and phase.To our investigation,this is the first time in HDS that multilevel complex amplitude demodulation is achieved experimentally from one diffraction intensity image without iterations.展开更多
Characterized by self-monitoring and agile adaptation to fast changing dynamics in complex production environments,smart manufacturing as envisioned under Industry 4.0 aims to improve the throughput and reliability of...Characterized by self-monitoring and agile adaptation to fast changing dynamics in complex production environments,smart manufacturing as envisioned under Industry 4.0 aims to improve the throughput and reliability of production beyond the state-of-the-art.While the widespread application of deep learning(DL)has opened up new opportunities to accomplish the goal,data quality and model interpretability have continued to present a roadblock for the widespread acceptance of DL for real-world applications.This has motivated research on two fronts:data curation,which aims to provide quality data as input for meaningful DL-based analysis,and model interpretation,which intends to reveal the physical reasoning underlying DL model outputs and promote trust from the users.This paper summarizes several key techniques in data curation where breakthroughs in data denoising,outlier detection,imputation,balancing,and semantic annotation have demonstrated the effectiveness in information extraction from noisy,incomplete,insufficient,and/or unannotated data.Also highlighted are model interpretation methods that address the“black-box”nature of DL towards model transparency.展开更多
1 Introduction On-device deep learning(DL)on mobile and embedded IoT devices drives various applications[1]like robotics image recognition[2]and drone swarm classification[3].Efficient local data processing preserves ...1 Introduction On-device deep learning(DL)on mobile and embedded IoT devices drives various applications[1]like robotics image recognition[2]and drone swarm classification[3].Efficient local data processing preserves privacy,enhances responsiveness,and saves bandwidth.However,current ondevice DL relies on predefined patterns,leading to accuracy and efficiency bottlenecks.It is difficult to provide feedback on data processing performance during the data acquisition stage,as processing typically occurs after data acquisition.展开更多
Low-field(nuclear magnetic resonance)NMR has been widely used in petroleum industry,such as well logging and laboratory rock core analysis.However,the signal-to-noise ratio is low due to the low magnetic field strengt...Low-field(nuclear magnetic resonance)NMR has been widely used in petroleum industry,such as well logging and laboratory rock core analysis.However,the signal-to-noise ratio is low due to the low magnetic field strength of NMR tools and the complex petrophysical properties of detected samples.Suppressing the noise and highlighting the available NMR signals is very important for subsequent data processing.Most denoising methods are normally based on fixed mathematical transformation or handdesign feature selectors to suppress noise characteristics,which may not perform well because of their non-adaptive performance to different noisy signals.In this paper,we proposed a“data processing framework”to improve the quality of low field NMR echo data based on dictionary learning.Dictionary learning is a machine learning method based on redundancy and sparse representation theory.Available information in noisy NMR echo data can be adaptively extracted and reconstructed by dictionary learning.The advantages and application effectiveness of the proposed method were verified with a number of numerical simulations,NMR core data analyses,and NMR logging data processing.The results show that dictionary learning can significantly improve the quality of NMR echo data with high noise level and effectively improve the accuracy and reliability of inversion results.展开更多
文摘Sentiment analysis, a crucial task in discerning emotional tones within the text, plays a pivotal role in understandingpublic opinion and user sentiment across diverse languages.While numerous scholars conduct sentiment analysisin widely spoken languages such as English, Chinese, Arabic, Roman Arabic, and more, we come to grapplingwith resource-poor languages like Urdu literature which becomes a challenge. Urdu is a uniquely crafted language,characterized by a script that amalgamates elements from diverse languages, including Arabic, Parsi, Pashtu,Turkish, Punjabi, Saraiki, and more. As Urdu literature, characterized by distinct character sets and linguisticfeatures, presents an additional hurdle due to the lack of accessible datasets, rendering sentiment analysis aformidable undertaking. The limited availability of resources has fueled increased interest among researchers,prompting a deeper exploration into Urdu sentiment analysis. This research is dedicated to Urdu languagesentiment analysis, employing sophisticated deep learning models on an extensive dataset categorized into fivelabels: Positive, Negative, Neutral, Mixed, and Ambiguous. The primary objective is to discern sentiments andemotions within the Urdu language, despite the absence of well-curated datasets. To tackle this challenge, theinitial step involves the creation of a comprehensive Urdu dataset by aggregating data from various sources such asnewspapers, articles, and socialmedia comments. Subsequent to this data collection, a thorough process of cleaningand preprocessing is implemented to ensure the quality of the data. The study leverages two well-known deeplearningmodels, namely Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN), for bothtraining and evaluating sentiment analysis performance. Additionally, the study explores hyperparameter tuning tooptimize the models’ efficacy. Evaluation metrics such as precision, recall, and the F1-score are employed to assessthe effectiveness of the models. The research findings reveal that RNN surpasses CNN in Urdu sentiment analysis,gaining a significantly higher accuracy rate of 91%. This result accentuates the exceptional performance of RNN,solidifying its status as a compelling option for conducting sentiment analysis tasks in the Urdu language.
基金supported by the National Natural Science Foundation of China(Grant Nos.42130719 and 42177173)the Doctoral Direct Train Project of Chongqing Natural Science Foundation(Grant No.CSTB2023NSCQ-BSX0029).
文摘Underground engineering projects such as deep tunnel excavation often encounter rockburst disasters accompanied by numerous microseismic events.Rapid interpretation of microseismic signals is crucial for the timely identification of rockbursts.However,conventional processing encompasses multi-step workflows,including classification,denoising,picking,locating,and computational analysis,coupled with manual intervention,which collectively compromise the reliability of early warnings.To address these challenges,this study innovatively proposes the“microseismic stethoscope"-a multi-task machine learning and deep learning model designed for the automated processing of massive microseismic signals.This model efficiently extracts three key parameters that are necessary for recognizing rockburst disasters:rupture location,microseismic energy,and moment magnitude.Specifically,the model extracts raw waveform features from three dedicated sub-networks:a classifier for source zone classification,and two regressors for microseismic energy and moment magnitude estimation.This model demonstrates superior efficiency compared to traditional processing and semi-automated processing,reducing per-event processing time from 0.71 s to 0.49 s to merely 0.036 s.It concurrently achieves 98%accuracy in source zone classification,with microseismic energy and moment magnitude estimation errors of 0.13 and 0.05,respectively.This model has been well applied and validated in the Daxiagu Tunnel case in Sichuan,China.The application results indicate that the model is as accurate as traditional methods in determining source parameters,and thus can be used to identify potential geomechanical processes of rockburst disasters.By enhancing the signal processing reliability of microseismic events,the proposed model in this study presents a significant advancement in the identification of rockburst disasters.
文摘The rapid growth of biomedical data,particularly multi-omics data including genomes,transcriptomics,proteomics,metabolomics,and epigenomics,medical research and clinical decision-making confront both new opportunities and obstacles.The huge and diversified nature of these datasets cannot always be managed using traditional data analysis methods.As a consequence,deep learning has emerged as a strong tool for analysing numerous omics data due to its ability to handle complex and non-linear relationships.This paper explores the fundamental concepts of deep learning and how they are used in multi-omics medical data mining.We demonstrate how autoencoders,variational autoencoders,multimodal models,attention mechanisms,transformers,and graph neural networks enable pattern analysis and recognition across all omics data.Deep learning has been found to be effective in illness classification,biomarker identification,gene network learning,and therapeutic efficacy prediction.We also consider critical problems like as data quality,model explainability,whether findings can be repeated,and computational power requirements.We now consider future elements of combining omics with clinical and imaging data,explainable AI,federated learning,and real-time diagnostics.Overall,this study emphasises the need of collaborating across disciplines to advance deep learning-based multi-omics research for precision medicine and comprehending complicated disorders.
基金supported by Princess Nourah bint Abdulrahman University Researchers Supporting Project number(PNURSP2026R765),Princess Nourah bint Abdulrahman University,Riyadh,Saudi Arabia.
文摘Human Activity Recognition(HAR)is a novel area for computer vision.It has a great impact on healthcare,smart environments,and surveillance while is able to automatically detect human behavior.It plays a vital role in many applications,such as smart home,healthcare,human computer interaction,sports analysis,and especially,intelligent surveillance.In this paper,we propose a robust and efficient HAR system by leveraging deep learning paradigms,including pre-trained models,CNN architectures,and their average-weighted fusion.However,due to the diversity of human actions and various environmental influences,as well as a lack of data and resources,achieving high recognition accuracy remain elusive.In this work,a weighted average ensemble technique is employed to fuse three deep learning models:EfficientNet,ResNet50,and a custom CNN.The results of this study indicate that using a weighted average ensemble strategy for developing more effective HAR models may be a promising idea for detection and classification of human activities.Experiments by using the benchmark dataset proved that the proposed weighted ensemble approach outperformed existing approaches in terms of accuracy and other key performance measures.The combined average-weighted ensemble of pre-trained and CNN models obtained an accuracy of 98%,compared to 97%,96%,and 95%for the customized CNN,EfficientNet,and ResNet50 models,respectively.
基金supported by the National Research Foundation of Korea(NRF)grant funded by theKorea government(MSIT)(No.RS-2024-00405278)partially supported by the Jeju Industry-University Convergence District Project for Promoting Industry-Campus Cooperationfunded by the Ministry of Trade,Industry and Energy(MOTIE,Korea)[Project Name:Jeju Industry-University Convergence District Project for Promoting Industry-Campus Cooperation/Project Number:P0029950].
文摘Recent advances in deep learning have significantly improved flood detection and segmentation from aerial and satellite imagery.However,conventional convolutional neural networks(CNNs)often struggle in complex flood scenarios involving reflections,occlusions,or indistinct boundaries due to limited contextual modeling.To address these challenges,we propose a hybrid flood segmentation framework that integrates a Vision Transformer(ViT)encoder with a U-Net decoder,enhanced by a novel Flood-Aware Refinement Block(FARB).The FARB module improves boundary delineation and suppresses noise by combining residual smoothing with spatial-channel attention mechanisms.We evaluate our model on a UAV-acquired flood imagery dataset,demonstrating that the proposed ViTUNet+FARB architecture outperforms existing CNN and Transformer-based models in terms of accuracy and mean Intersection over Union(mIoU).Detailed ablation studies further validate the contribution of each component,confirming that the FARB design significantly enhances segmentation quality.To its better performance and computational efficiency,the proposed framework is well-suited for flood monitoring and disaster response applications,particularly in resource-constrained environments.
基金supported by the National Natural Science Foundation of China(Nos.62171469,62071029)。
文摘The estimation of the Number of Sources(NoS)is a significant challenge in signal processing,particularly due to the impact of colored noise on the performance of NoS estimation.This paper proposes a Multidimensional Feature Network(MFNet)which is designed for NoS estimation by extracting features of the sampled received signals and Sampled Covariance Matrix(SCM).The MFNet treats the raw signal and the SCM as two different types of data,and is able to achieve NoS estimation under colored noise and imperfect array.MFNet employs the Gated Recurrent Unit(GRU)to capture sequential information from the original signal data and to construct the Pseudo Covariance Matrix(PCM).Subsequently,various dimensional features,including eigenvalues and the Gerschgorin disk radius,are extracted from both the PCM and SCM,which are then jointly input into the subsequent network.An overall accuracy of 82%can be achieved after network training.The ablation experimental results demonstrate the effectiveness of multiple inputs.And simulation results demonstrate that the proposed MFNet achieves higher estimation accuracy compared to existing algorithms and exhibits greater robustness against colored noise.
基金the National Science and Technology Council,Taiwan,for financially supporting this research(grant No.NSTC 114-2221-E-018-003)the Ministry of Education’s Teaching Practice Research Program,Taiwan(PSK1142780).
文摘This study focuses on developing a deep learning model capable of recognizing vehicle brands and models,integrated with a law enforcement intelligence platform to overcome the limitations of existing license plate recognition techniques—particularly in handling counterfeit,obscured,or absent plates.The research first entailed collecting,annotating,and classifying images of various vehiclemodels,leveraging image processing and feature extraction methodologies to train themodel on Microsoft Custom Vision.Experimental results indicate that,formost brands and models,the system achieves stable and relatively high performance in Precision,Recall,and Average Precision(AP).Furthermore,simulated tests involving illicit vehicles reveal that,even in cases of reassigned,concealed,or missing license plates,the model can rely on exterior body features to effectively identify vehicles,reducing dependence on plate-specific data.In practical law enforcement scenarios,these findings can accelerate investigations of stolen or forged plates and enhance overall accuracy.In conclusion,continued collection of vehicle images across broadermodel types,production years,and modification levels—along with refined annotation processes and parameter adjustment strategies—will further strengthen themethod’s applicability within law enforcement intelligence platforms,facilitating more precise and comprehensive vehicle recognition and control in real-world operations.
基金supported by grants from the National Key Research and Development Program of China(2022YFD2001103)the National Natural Science Foundation of China(42371373)。
文摘Near real-time maize phenology monitoring is crucial for field management,cropping system adjustments,and yield estimation.Most phenological monitoring methods are post-seasonal and heavily rely on high-frequency time-series data.These methods are not applicable on the unmanned aerial vehicle(UAV)platform due to the high cost of acquiring time-series UAV images and the shortage of UAV-based phenological monitoring methods.To address these challenges,we employed the Synthetic Minority Oversampling Technique(SMOTE)for sample augmentation,aiming to resolve the small sample modelling problem.Moreover,we utilized enhanced"separation"and"compactness"feature selection methods to identify input features from multiple data sources.In this process,we incorporated dynamic multi-source data fusion strategies,involving Vegetation index(VI),Color index(CI),and Texture features(TF).A two-stage neural network that combines Convolutional Neural Network(CNN)and Long Short-Term Memory Network(LSTM)is proposed to identify maize phenological stages(including sowing,seedling,jointing,trumpet,tasseling,maturity,and harvesting)on UAV platforms.The results indicate that the dataset generated by SMOTE closely resembles the measured dataset.Among dynamic data fusion strategies,the VI-TF combination proves to be most effective,with CI-TF and VI-CI combinations following behind.Notably,as more data sources are integrated,the model's demand for input features experiences a significant decline.In particular,the CNN-LSTM model,based on the fusion of three data sources,exhibited remarkable reliability when validating the three datasets.For Dataset 1(Beijing Xiaotangshan,2023:Data from 12 UAV Flight Missions),the model achieved an overall accuracy(OA)of 86.53%.Additionally,its precision(Pre),recall(Rec),F1 score(F1),false acceptance rate(FAR),and false rejection rate(FRR)were 0.89,0.89,0.87,0.11,and 0.11,respectively.The model also showed strong generalizability in Dataset 2(Beijing Xiaotangshan,2023:Data from 6 UAV Flight Missions)and Dataset 3(Beijing Xiaotangshan,2022:Data from 4 UAV Flight Missions),with OAs of 89.4%and 85%,respectively.Meanwhile,the model has a low demand for input featu res,requiring only 54.55%(99 of all featu res).The findings of this study not only offer novel insights into near real-time crop phenology monitoring,but also provide technical support for agricultural field management and cropping system adaptation.
基金supported by the PetroChina Prospective,Basic,and Strategic Technology Research Project(No.2021ZG03-02 and No.2023DJ8402)。
文摘The deep learning algorithm,which has been increasingly applied in the field of petroleum geophysical prospecting,has achieved good results in improving efficiency and accuracy based on test applications.To play a greater role in actual production,these algorithm modules must be integrated into software systems and used more often in actual production projects.Deep learning frameworks,such as TensorFlow and PyTorch,basically take Python as the core architecture,while the application program mainly uses Java,C#,and other programming languages.During integration,the seismic data read by the Java and C#data interfaces must be transferred to the Python main program module.The data exchange methods between Java,C#,and Python include shared memory,shared directory,and so on.However,these methods have the disadvantages of low transmission efficiency and unsuitability for asynchronous networks.Considering the large volume of seismic data and the need for network support for deep learning,this paper proposes a method of transmitting seismic data based on Socket.By maximizing Socket’s cross-network and efficient longdistance transmission,this approach solves the problem of inefficient transmission of underlying data while integrating the deep learning algorithm module into a software system.Furthermore,the actual production application shows that this method effectively solves the shortage of data transmission in shared memory,shared directory,and other modes while simultaneously improving the transmission efficiency of massive seismic data across modules at the bottom of the software.
基金supported in part by the Science and Technology Department of Sichuan Province(No.2025ZNSFSC0427,No.2024ZDZX0035)the Open Project Fund of Vehicle Measurement,Control and Safety Key Laboratory of Sichuan Province(No.QCCK2024-004)the Industrial and Educational Integration Project of Yibin(No.YB-XHU-20240001)。
文摘The accurate prediction of battery pack capacity in electric vehicles(EVs)is crucial for ensuring safety and optimizing performance.Despite extensive research on predicting cell capacity using laboratory data,predicting the capacity of onboard battery packs from field data remains challenging due to complex operating conditions and irregular EV usage in real-world settings.Most existing methods rely on extracting health feature parameters from raw data for capacity prediction of onboard battery packs,however,selecting specific parameters often results in a loss of critical information,which reduces prediction accuracy.To this end,this paper introduces a novel framework combining deep learning and data compression techniques to accurately predict battery pack capacity onboard.The proposed data compression method converts monthly EV charging data into feature maps,which preserve essential data characteristics while reducing the volume of raw data.To address missing capacity labels in field data,a capacity labeling method is proposed,which calculates monthly battery capacity by transforming the ampere-hour integration formula and applying linear regression.Subsequently,a deep learning model is proposed to build a capacity prediction model,using feature maps from historical months to predict the battery capacity of future months,thus facilitating accurate forecasts.The proposed framework,evaluated using field data from 20 EVs,achieves a mean absolute error of 0.79 Ah,a mean absolute percentage error of 0.65%,and a root mean square error of 1.02 Ah,highlighting its potential for real-world EV applications.
基金funded by the Researchers Supporting Project number(RSPD2025R857),King Saud University,Riyadh,Saudi Arabia.
文摘Many bioinformatics applications require determining the class of a newly sequenced Deoxyribonucleic acid(DNA)sequence,making DNA sequence classification an integral step in performing bioinformatics analysis,where large biomedical datasets are transformed into valuable knowledge.Existing methods rely on a feature extraction step and suffer from high computational time requirements.In contrast,newer approaches leveraging deep learning have shown significant promise in enhancing accuracy and efficiency.In this paper,we investigate the performance of various deep learning architectures:Convolutional Neural Network(CNN),CNN-Long Short-Term Memory(CNNLSTM),CNN-Bidirectional Long Short-Term Memory(CNN-BiLSTM),Residual Network(ResNet),and InceptionV3 for DNA sequence classification.Various numerical and visual data representation techniques are utilized to represent the input datasets,including:label encoding,k-mer sentence encoding,k-mer one-hot vector,Frequency Chaos Game Representation(FCGR)and 5-Color Map(ColorSquare).Three datasets are used for the training of the models including H3,H4 and DNA Sequence Dataset(Yeast,Human,Arabidopsis Thaliana).Experiments are performed to determine which combination of DNA representation and deep learning architecture yields improved performance for the classification task.Our results indicate that using a hybrid CNN-LSTM neural network trained on DNA sequences represented as one-hot encoded k-mer sequences yields the best performance,achieving an accuracy of 92.1%.
基金supported by the National Natural Science Foundation of China[Grant No.:82172524]the Natural Science Foundation of Hubei Province[Grant No.:2025AFB240].
文摘Bone tumors(BTs)-including osteosarcoma,Ewing sarcoma,and chondrosarcoma-are rare but biologically complex malignancies characterized by pronounced heterogeneity in anatomical location,histological subtype,and molecular alterations.Recent advances in artificial intelligence(AI),particularly deep learning,have enabled the integration of diverse clinical data modalities to support diagnosis,treatment planning,and prognostication in bone oncology.This review provides a comprehensive synthesis of AI-driven multimodal fusion strategies that incorporate radiological imaging,digital pathology,multi-omics profiling,and electronic health records.We conducted a structured review of peer-reviewed literature published between 2015 and early 2025,focusing on the development,validation,and clinical applicability of AI models for BT diagnosis,subtyping,treatment response prediction,and recurrence monitoring.Although multimodal models have demonstrated advantages over unimodal approaches,especially in handling missing data and improving generalizability,most remain constrained by single-center study designs,small sample sizes,and limited prospective or external validation.Persistent technical and translational challenges include semantic misalignment across modalities,incomplete datasets,limited model interpretability,and regulatory and infrastructural barriers to clinical integration.To address these limitations,we highlight emerging directions such as contrastive representation learning,generative data augmentation,transformer-based fusion architectures,and privacy-preserving federated learning.We also discuss the evolving role of foundation models and workflow-integrated AI agents in enhancing scalability and clinical usability.In summary,multimodal AI represents a promising paradigm for advancing precision care in BTs.Realizing its full clinical potential will require methodologically rigorous,biologically informed,and system-level approaches that bridge algorithmic innovation with real-world healthcare delivery.
基金The National Natural Science Foundation of China under contract Nos 42176011 and 61931025the Fundamental Research Funds for the Central Universities of China under contract No.24CX03001A.
文摘Efficient and accurate prediction of ocean surface latent heat fluxes is essential for understanding and modeling climate dynamics.Conventional estimation methods have low resolution and lack accuracy.The transformer model,with its self-attention mechanism,effectively captures long-range dependencies,leading to a degradation of accuracy over time.Due to the non-linearity and uncertainty of physical processes,the transformer model encounters the problem of error accumulation,leading to a degradation of accuracy over time.To solve this problem,we combine the Data Assimilation(DA)technique with the transformer model and continuously modify the model state to make it closer to the actual observations.In this paper,we propose a deep learning model called TransNetDA,which integrates transformer,convolutional neural network and DA methods.By combining data-driven and DA methods for spatiotemporal prediction,TransNetDA effectively extracts multi-scale spatial features and significantly improves prediction accuracy.The experimental results indicate that the TransNetDA method surpasses traditional techniques in terms of root mean square error and R2 metrics,showcasing its superior performance in predicting latent heat fluxes at the ocean surface.
基金supported by the Science Committee of the Ministry of Higher Education and Science of the Republic of Kazakhstan within the framework of grant AP23489899“Applying Deep Learning and Neuroimaging Methods for Brain Stroke Diagnosis”.
文摘Deep learning now underpins many state-of-the-art systems for biomedical image and signal processing,enabling automated lesion detection,physiological monitoring,and therapy planning with accuracy that rivals expert performance.This survey reviews the principal model families as convolutional,recurrent,generative,reinforcement,autoencoder,and transfer-learning approaches as emphasising how their architectural choices map to tasks such as segmentation,classification,reconstruction,and anomaly detection.A dedicated treatment of multimodal fusion networks shows how imaging features can be integrated with genomic profiles and clinical records to yield more robust,context-aware predictions.To support clinical adoption,we outline post-hoc explainability techniques(Grad-CAM,SHAP,LIME)and describe emerging intrinsically interpretable designs that expose decision logic to end users.Regulatory guidance from the U.S.FDA,the European Medicines Agency,and the EU AI Act is summarised,linking transparency and lifecycle-monitoring requirements to concrete development practices.Remaining challenges as data imbalance,computational cost,privacy constraints,and cross-domain generalization are discussed alongside promising solutions such as federated learning,uncertainty quantification,and lightweight 3-D architectures.The article therefore offers researchers,clinicians,and policymakers a concise,practice-oriented roadmap for deploying trustworthy deep-learning systems in healthcare.
文摘Crop phenotyping plays a critical role in precision agriculture by enabling the accurate assessment of plant traits,supporting improved crop management,breeding programs,and yield optimization.However,cowpea leaves present unique challenges for automated phenotyping due to their diverse shapes,complex vein structures,and variations caused by environmental conditions.This research presents a deep learning-based approach for the classification of cowpea leaf images to support crop phenotyping tasks.Given the limited availability of annotated datasets,data augmentation techniques were employed to artificially expand the original small dataset while preserving essential leaf characteristics.Various image processing methods were applied to enrich the dataset,ensuring better feature representation without significant information loss.A deep neural network,specifically the MobileNet architecture,was utilized for its efficiency in capturing multi-scale features and handling image data with limited computational resources.The performance of the model trained on the augmented dataset was evaluated,achieving an accuracy of 94.12%on the cowpea leaf classification task.These results demonstrate the effectiveness of data augmentation in enhancing model generalization and learning capabilities.
文摘Lung cancer continues to be a leading cause of cancer-related deaths worldwide,emphasizing the critical need for improved diagnostic techniques.Early detection of lung tumors significantly increases the chances of successful treatment and survival.However,current diagnostic methods often fail to detect tumors at an early stage or to accurately pinpoint their location within the lung tissue.Single-model deep learning technologies for lung cancer detection,while beneficial,cannot capture the full range of features present in medical imaging data,leading to incomplete or inaccurate detection.Furthermore,it may not be robust enough to handle the wide variability in medical images due to different imaging conditions,patient anatomy,and tumor characteristics.To overcome these disadvantages,dual-model or multi-model approaches can be employed.This research focuses on enhancing the detection of lung cancer by utilizing a combination of two learning models:a Convolutional Neural Network(CNN)for categorization and the You Only Look Once(YOLOv8)architecture for real-time identification and pinpointing of tumors.CNNs automatically learn to extract hierarchical features from raw image data,capturing patterns such as edges,textures,and complex structures that are crucial for identifying lung cancer.YOLOv8 incorporates multiscale feature extraction,enabling the detection of tumors of varying sizes and scales within a single image.This is particularly beneficial for identifying small or irregularly shaped tumors that may be challenging to detect.Furthermore,through the utilization of cutting-edge data augmentation methods,such as Deep Convolutional Generative Adversarial Networks(DCGAN),the suggested approach can handle the issue of limited data and boost the models’ability to learn from diverse and comprehensive datasets.The combined method not only improved accuracy and localization but also ensured efficient real-time processing,which is crucial for practical clinical applications.The CNN achieved an accuracy of 97.67%in classifying lung tissues into healthy and cancerous categories.The YOLOv8 model achieved an Intersection over Union(IoU)score of 0.85 for tumor localization,reflecting high precision in detecting and marking tumor boundaries within the images.Finally,the incorporation of synthetic images generated by DCGAN led to a 10%improvement in both the CNN classification accuracy and YOLOv8 detection performance.
基金We are grateful for financial supports from National Key Research and Development Program of China(2018YFA0701800)Project of Fujian Province Major Science and Technology(2020HZ01012)+1 种基金Natural Science Foundation of Fujian Province(2021J01160)National Natural Science Foundation of China(62061136005).
文摘To increase the storage capacity in holographic data storage(HDS),the information to be stored is encoded into a complex amplitude.Fast and accurate retrieval of amplitude and phase from the reconstructed beam is necessary during data readout in HDS.In this study,we proposed a complex amplitude demodulation method based on deep learning from a single-shot diffraction intensity image and verified it by a non-interferometric lensless experiment demodulating four-level amplitude and four-level phase.By analyzing the correlation between the diffraction intensity features and the amplitude and phase encoding data pages,the inverse problem was decomposed into two backward operators denoted by two convolutional neural networks(CNNs)to demodulate amplitude and phase respectively.The experimental system is simple,stable,and robust,and it only needs a single diffraction image to realize the direct demodulation of both amplitude and phase.To our investigation,this is the first time in HDS that multilevel complex amplitude demodulation is achieved experimentally from one diffraction intensity image without iterations.
文摘Characterized by self-monitoring and agile adaptation to fast changing dynamics in complex production environments,smart manufacturing as envisioned under Industry 4.0 aims to improve the throughput and reliability of production beyond the state-of-the-art.While the widespread application of deep learning(DL)has opened up new opportunities to accomplish the goal,data quality and model interpretability have continued to present a roadblock for the widespread acceptance of DL for real-world applications.This has motivated research on two fronts:data curation,which aims to provide quality data as input for meaningful DL-based analysis,and model interpretation,which intends to reveal the physical reasoning underlying DL model outputs and promote trust from the users.This paper summarizes several key techniques in data curation where breakthroughs in data denoising,outlier detection,imputation,balancing,and semantic annotation have demonstrated the effectiveness in information extraction from noisy,incomplete,insufficient,and/or unannotated data.Also highlighted are model interpretation methods that address the“black-box”nature of DL towards model transparency.
基金supported by the National Science Fund for Distinguished Young Scholars(62025205)the National Natural Science Foundation of China(Grant Nos.62032020,62102317)CityU APRC Grant(9610633).
文摘1 Introduction On-device deep learning(DL)on mobile and embedded IoT devices drives various applications[1]like robotics image recognition[2]and drone swarm classification[3].Efficient local data processing preserves privacy,enhances responsiveness,and saves bandwidth.However,current ondevice DL relies on predefined patterns,leading to accuracy and efficiency bottlenecks.It is difficult to provide feedback on data processing performance during the data acquisition stage,as processing typically occurs after data acquisition.
基金supported by Science Foundation of China University of Petroleum,Beijing(Grant Number ZX20210024)Chinese Postdoctoral Science Foundation(Grant Number 2021M700172)+1 种基金The Strategic Cooperation Technology Projects of CNPC and CUP(Grant Number ZLZX2020-03)National Natural Science Foundation of China(Grant Number 42004105)
文摘Low-field(nuclear magnetic resonance)NMR has been widely used in petroleum industry,such as well logging and laboratory rock core analysis.However,the signal-to-noise ratio is low due to the low magnetic field strength of NMR tools and the complex petrophysical properties of detected samples.Suppressing the noise and highlighting the available NMR signals is very important for subsequent data processing.Most denoising methods are normally based on fixed mathematical transformation or handdesign feature selectors to suppress noise characteristics,which may not perform well because of their non-adaptive performance to different noisy signals.In this paper,we proposed a“data processing framework”to improve the quality of low field NMR echo data based on dictionary learning.Dictionary learning is a machine learning method based on redundancy and sparse representation theory.Available information in noisy NMR echo data can be adaptively extracted and reconstructed by dictionary learning.The advantages and application effectiveness of the proposed method were verified with a number of numerical simulations,NMR core data analyses,and NMR logging data processing.The results show that dictionary learning can significantly improve the quality of NMR echo data with high noise level and effectively improve the accuracy and reliability of inversion results.
