Remote sensing cross-modal image-text retrieval(RSCIR)can flexibly and subjectively retrieve remote sensing images utilizing query text,which has received more researchers’attention recently.However,with the increasi...Remote sensing cross-modal image-text retrieval(RSCIR)can flexibly and subjectively retrieve remote sensing images utilizing query text,which has received more researchers’attention recently.However,with the increasing volume of visual-language pre-training model parameters,direct transfer learning consumes a substantial amount of computational and storage resources.Moreover,recently proposed parameter-efficient transfer learning methods mainly focus on the reconstruction of channel features,ignoring the spatial features which are vital for modeling key entity relationships.To address these issues,we design an efficient transfer learning framework for RSCIR,which is based on spatial feature efficient reconstruction(SPER).A concise and efficient spatial adapter is introduced to enhance the extraction of spatial relationships.The spatial adapter is able to spatially reconstruct the features in the backbone with few parameters while incorporating the prior information from the channel dimension.We conduct quantitative and qualitative experiments on two different commonly used RSCIR datasets.Compared with traditional methods,our approach achieves an improvement of 3%-11% in sumR metric.Compared with methods finetuning all parameters,our proposed method only trains less than 1% of the parameters,while maintaining an overall performance of about 96%.展开更多
Research on mass gathering events is critical for ensuring public security and maintaining social order.However,most of the existing works focus on crowd behavior analysis areas such as anomaly detection and crowd cou...Research on mass gathering events is critical for ensuring public security and maintaining social order.However,most of the existing works focus on crowd behavior analysis areas such as anomaly detection and crowd counting,and there is a relative lack of research on mass gathering behaviors.We believe real-time detection and monitoring of mass gathering behaviors are essential formigrating potential security risks and emergencies.Therefore,it is imperative to develop a method capable of accurately identifying and localizing mass gatherings before disasters occur,enabling prompt and effective responses.To address this problem,we propose an innovative Event-Driven Attention Network(EDAN),which achieves image-text matching in the scenario of mass gathering events with good results for the first time.Traditional image-text retrieval methods based on global alignment are difficult to capture the local details within complex scenes,limiting retrieval accuracy.While local alignment-based methods aremore effective at extracting detailed features,they frequently process raw textual features directly,which often contain ambiguities and redundant information that can diminish retrieval efficiency and degrade model performance.To overcome these challenges,EDAN introduces an Event-Driven AttentionModule that adaptively focuses attention on image regions or textual words relevant to the event type.By calculating the semantic distance between event labels and textual content,this module effectively significantly reduces computational complexity and enhances retrieval efficiency.To validate the effectiveness of EDAN,we construct a dedicated multimodal dataset tailored for the analysis of mass gathering events,providing a reliable foundation for subsequent studies.We conduct comparative experiments with other methods on our dataset,the experimental results demonstrate the effectiveness of EDAN.In the image-to-text retrieval task,EDAN achieved the best performance on the R@5 metric,while in the text-to-image retrieval task,it showed superior results on both R@10 and R@5 metrics.Additionally,EDAN excelled in the overall Rsummetric,achieving the best performance.Finally,ablation studies further verified the effectiveness of event-driven attention module.展开更多
With the incessant propulsion of the Open Door Policy,which is related to the consolidation of international collaborative partnerships,an increasing number of Chinese companies are moving toward cooperating countries...With the incessant propulsion of the Open Door Policy,which is related to the consolidation of international collaborative partnerships,an increasing number of Chinese companies are moving toward cooperating countries to participate in infrastructure construction,employing a win-win strategy in favor of the people and governments of both countries.Among the cooperation domains,our country’s electrical companies have achieved a series of remarkable results in the international Engineering,Procurement,and Construction(EPC)project market with their outstanding business capabilities and technical advantages.Nevertheless,some shortcomings cannot be overlooked,the most notable of which appears to be the impediment associated with engineering translation,which has always been an obsession among translators of Chinese companies.Taking the transmission line project in the Republic of Madagascar as an example,an analysis of French-Chinese translation methods of electrical engineering terminology in the field of the transmission line is carried out.展开更多
In recent years,cross-modal hash retrieval has become a popular research field because of its advantages of high efficiency and low storage.Cross-modal retrieval technology can be applied to search engines,crossmodalm...In recent years,cross-modal hash retrieval has become a popular research field because of its advantages of high efficiency and low storage.Cross-modal retrieval technology can be applied to search engines,crossmodalmedical processing,etc.The existing main method is to use amulti-label matching paradigm to finish the retrieval tasks.However,such methods do not use fine-grained information in the multi-modal data,which may lead to suboptimal results.To avoid cross-modal matching turning into label matching,this paper proposes an end-to-end fine-grained cross-modal hash retrieval method,which can focus more on the fine-grained semantic information of multi-modal data.First,the method refines the image features and no longer uses multiple labels to represent text features but uses BERT for processing.Second,this method uses the inference capabilities of the transformer encoder to generate global fine-grained features.Finally,in order to better judge the effect of the fine-grained model,this paper uses the datasets in the image text matching field instead of the traditional label-matching datasets.This article experiment on Microsoft COCO(MS-COCO)and Flickr30K datasets and compare it with the previous classicalmethods.The experimental results show that this method can obtain more advanced results in the cross-modal hash retrieval field.展开更多
The arms race between avian brood parasites and their hosts provides a classic model for studying coevolution.In one of the most widespread obligate brood parasites,the Common Cuckoo(Cuculus canorus),chicks typically ...The arms race between avian brood parasites and their hosts provides a classic model for studying coevolution.In one of the most widespread obligate brood parasites,the Common Cuckoo(Cuculus canorus),chicks typically evict all host progeny(eggs and nestlings)from the nest cup,resulting in complete reproductive failure for the host.Host parents of Common Cuckoos could thus potentially benefit from retrieving evicted eggs and nestlings into the nest cup.However,whether hosts of the Common Cuckoo exhibit such retrieval behavior has been scarcely studied.In this study,we experimentally investigated the occurrence of retrieval in a nestbox-breeding population of Daurian Redstarts(Phoenicurus auroreus),a common cavity-nesting host of the Common Cuckoo.To test the redstarts'response to an egg or a nestling outside the nest cup,we experimentally placed either a conspecific egg,a model cuckoo egg,or a redstart nestling near the rim of the nest cup.We found that redstarts never showed retrieval behavior of either eggs or nestlings.All hosts ignored the experimental nestling and conspecific egg,but most ejected the model cuckoo egg from the nestbox.Our results suggest that selection for retrieval behavior in this cavity-nesting host may be weak or even negative.We discuss several ecological and evolutionary factors that may explain the absence of retrieval in this system.展开更多
Accurate retrieval of atmospheric vertical profiles is critical for improving weather prediction and climate monitoring.However,the complexity of atmospheric processes in cloudy regions poses challenges compared to th...Accurate retrieval of atmospheric vertical profiles is critical for improving weather prediction and climate monitoring.However,the complexity of atmospheric processes in cloudy regions poses challenges compared to those of clear sky scenarios.This study presents a novel framework that integrates Bayesian optimization and machine learning approaches to retrieve atmospheric vertical profiles—including temperature,humidity,ozone concentration,cloud fraction,ice water content(IWC),and liquid water content(LWC)—from hyperspectral infrared observations.Specifically,a Bayesian method was used to refine ERA5 reanalysis data by minimizing brightness temperature(BT)discrepancies against FY-4B Geostationary Interferometric Infrared Sounder(GIIRS)observations,generating a high-quality profile database(~2.8 million profiles)across diverse weather systems.The optimized profiles improve radiative consistency,reducing BT biases from>40 K to<10 K in cloudy regions.To further overcome the limitations of the Bayesian method,we developed a Transformer-Resnet hybrid model(TERNet),which achieved superior performance with RMSE values of 1.61 K(temperature),5.77%(humidity),and 2.25×10^(–6)/6.09×10^(–6)kg kg^(–1)(IWC/LWC)across the entire vertical levels in all-sky conditions.The TERNet outperforms both ERA5 in cloud parameter retrieval and the GIIRS L2 product in thermodynamic profiling.Independent verification with radiosonde and Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations(CALIPSO)datasets confirms the framework's reliability across various meteorological regimes.This work demonstrates the capability of combining physics-informed Bayesian methods with data-driven machine learning to fully exploit hyperspectral IR data.展开更多
Background:Hormonal treatment and response as a predictor of sperm retrieval prior to microdissection testicular sperm extraction(micro-TESE)are not well established in the current literature.This study aimed to inves...Background:Hormonal treatment and response as a predictor of sperm retrieval prior to microdissection testicular sperm extraction(micro-TESE)are not well established in the current literature.This study aimed to investigate the hormonal response as a predictor of sperm retrieval among men with nonobstructive azoospermia(NOA).Methods:Seventy-seven consecutive patients who had testosterone levels≤14 nmol/L were treated medically with an aromatase inhibitor or recombinant human chorionic gonadotropin(rec-hCG)prior to micro-TESE and were included.Thirty-four(44.2%)had unexplained NOA(UNEX),25(32.5%)had Klinefelter syndrome(KS),8(10.4%)had a history of cryptorchidism(UDT),4(5.2%)had microdeletion of the Azoospermia factor C(AZFc),and 6(7.8%)were treated previously with chemotherapy.Baseline and post-treatment serum hormonal levels were documented.Pre-op testosterone levels were entered into binary logistic regressions with age,Follicle-stimulating hormone(FSH),and Luteinizing hormone(LH)levels to test for significance with sperm retrieval.We then built logistic regression models to identify predictors of successful surgical sperm retrieval(SSR).Results:Forty-five patients(58%)had successful retrieval.In 32 patients(42%),no sperm was retrieved.Both the mean pre-op testosterone and the mean testosterone change between the two groups were significant(p=0.02 and p=0.011,respectively).Receiver operating characteristic(ROC)analysis demonstrated an area under the curve(AUC)of 0.785(95%CI=0.685-0.886,p<0.001).The Youden index coefficient was calculated for KS and UNEX.The cut-off point for KS was established at 0.764(sensitivity=0.875,false positive rate[FPR]=0.111),and 0.215 for UNEX(sensitivity=0.438,FPR=0.222).We also observed a correlation between age and SSR(p=0.05).In KS patients,SSR was determined by pre-op testosterone levels irrespective of age.Conclusion:Pre-operative hormonal response is a predictor for SSR in NOA patients who were treated medically.This data may help during pre-operative counselling.展开更多
Cross-modal image-text retrieval is a fundamental task in bridging vision and language. It faces two main challenges that are typically not well addressed in previous works. 1) Generalizability: Existing methods often...Cross-modal image-text retrieval is a fundamental task in bridging vision and language. It faces two main challenges that are typically not well addressed in previous works. 1) Generalizability: Existing methods often assume a strong semantic correlation between each text-image pair, which are thus difficult to generalize to real-world scenarios where the weak correlation dominates. 2) Efficiency: Many latest works adopt the single-tower architecture with heavy detectors, which are inefficient during the inference stage because the costly computation needs to be repeated for each text-image pair. In this work, to overcome these two challenges, we propose a two-tower cross-modal contrastive learning (CMCL) framework. Specifically, we first devise a two-tower architecture, which enables a unified feature space for the text and image modalities to be directly compared with each other, alleviating the heavy computation during inference. We further introduce a simple yet effective module named multi-grid split (MGS) to learn fine-grained image features without using detectors. Last but not the least, we deploy a cross-modal contrastive loss on the global image/text features to learn their weak correlation and thus achieve high generalizability. To validate that our CMCL can be readily generalized to real-world scenarios, we construct a large multi-source image-text dataset called weak semantic correlation dataset (WSCD). Extensive experiments show that our CMCL outperforms the state-of-the-arts while being much more efficient.展开更多
Image-text retrieval aims to capture the semantic correspondence between images and texts,which serves as a foundation and crucial component in multi-modal recommendations,search systems,and online shopping.Existing m...Image-text retrieval aims to capture the semantic correspondence between images and texts,which serves as a foundation and crucial component in multi-modal recommendations,search systems,and online shopping.Existing mainstream methods primarily focus on modeling the association of image-text pairs while neglecting the advantageous impact of multi-task learning on image-text retrieval.To this end,a multi-task visual semantic embedding network(MVSEN)is proposed for image-text retrieval.Specifically,we design two auxiliary tasks,including text-text matching and multi-label classification,for semantic constraints to improve the generalization and robustness of visual semantic embedding from a training perspective.Besides,we present an intra-and inter-modality interaction scheme to learn discriminative visual and textual feature representations by facilitating information flow within and between modalities.Subsequently,we utilize multi-layer graph convolutional networks in a cascading manner to infer the correlation of image-text pairs.Experimental results show that MVSEN outperforms state-of-the-art methods on two publicly available datasets,Flickr30K and MSCOCO,with rSum improvements of 8.2%and 3.0%,respectively.展开更多
Image-based similar trademark retrieval is a time-consuming and labor-intensive task in the trademark examination process.This paper aims to support trademark examiners by training Deep Convolutional Neural Network(DC...Image-based similar trademark retrieval is a time-consuming and labor-intensive task in the trademark examination process.This paper aims to support trademark examiners by training Deep Convolutional Neural Network(DCNN)models for effective Trademark Image Retrieval(TIR).To achieve this goal,we first develop a novel labeling method that automatically generates hundreds of thousands of labeled similar and dissimilar trademark image pairs using accompanying data fields such as citation lists,Vienna classification(VC)codes,and trademark ownership information.This approach eliminates the need for manual labeling and provides a large-scale dataset suitable for training deep learning models.We then train DCNN models based on Siamese and Triplet architectures,evaluating various feature extractors to determine the most effective configuration.Furthermore,we present an Adapted Contrastive Loss Function(ACLF)for the trademark retrieval task,specifically engineered to mitigate the influence of noisy labels found in automatically created datasets.Experimental results indicate that our proposed model(Efficient-Net_v21_Siamese)performs best at both True Negative Rate(TNR)threshold levels,TNR 0.9 and TNR 0.95,with==respective True Positive Rates(TPRs)of 77.7%and 70.8%and accuracies of 83.9%and 80.4%.Additionally,when testing on the public trademark dataset METU_v2,our model achieves a normalized average rank(NAR)of 0.0169,outperforming the current state-of-the-art(SOTA)model.Based on these findings,we estimate that considering only approximately 10%of the returned trademarks would be sufficient,significantly reducing the review time.Therefore,the paper highlights the potential of utilizing national trademark data to enhance the accuracy and efficiency of trademark retrieval systems,ultimately supporting trademark examiners in their evaluation tasks.展开更多
This study describes the use of the weighted multiplicative algebraic reconstruction technique(WMART)to obtain vertical ozone profiles from limb observations performed by the scanning imaging absorption spectrometer f...This study describes the use of the weighted multiplicative algebraic reconstruction technique(WMART)to obtain vertical ozone profiles from limb observations performed by the scanning imaging absorption spectrometer for atmospheric chartography(SCIAMACHY).This technique is based on SaskMART(the combination of the multiplicative algebraic reconstruction technique and SaskTRAN radiative transfer model),which was originally developed for optical spectrometer and infrared imaging system(OSIRIS)data.One of the objectives of this study was to obtain consistent ozone profiles from the two satellites.In this study,the WMART algorithm is combined with a radiative transfer model(SCIATRAN),as well as a set of measurement vectors comprising five Hartley pairing vectors(HPVs)and one Chappuis triplet vector(CTV),to retrieve ozone profiles in the altitude range of 10–69 km.Considering that the weighting factors in WMART have a significant effect on the retrievals,we propose a novel approach to calculate the pair/triplet weighting factors using wavelength weighting functions.The results of the application of the proposed ozone retrieval scheme are compared with the SCIAMACHY v3.5 ozone product by University of Bremen and validated against profiles derived from other passive satellite observations or measured by ozonesondes.Between 18 and 55 km,the retrieved ozone profiles typically agree with data from the SCIAMACHY ozone product within 5%for tropics and middle latitudes,whereas a negative deviation exists between 35 and 50 km for northern high latitudes,with a deviation of less than 10%above 50 km.Comparison of the retrieved profiles with microwave limb sounder(MLS)v5.0 indicates that the difference is within±5%between 18 and 55 km,and an agreement within 10%is achieved in other altitudes for tropics and middle latitudes.Comparison of the retrieved profiles with OSIRIS v7.1 indicates that the average deviation is within±5%between 20 and 59 km,and difference of approximately 10%is achieved below 20 km.Compared with ozonesondes data,a general validity of the retrievals is no more than 5%between 15 and 30 km.展开更多
DDeeaarr EEddiittoorr,,The encoding and retrieval of emotional memories demands intricate interplay within the limbic network,where the network state is subject to significant reconfiguration by learning-induced plast...DDeeaarr EEddiittoorr,,The encoding and retrieval of emotional memories demands intricate interplay within the limbic network,where the network state is subject to significant reconfiguration by learning-induced plasticity,behavioral state,and contextual information[1].展开更多
Content-Based Image Retrieval(CBIR)and image mining are becoming more important study fields in computer vision due to their wide range of applications in healthcare,security,and various domains.The image retrieval sy...Content-Based Image Retrieval(CBIR)and image mining are becoming more important study fields in computer vision due to their wide range of applications in healthcare,security,and various domains.The image retrieval system mainly relies on the efficiency and accuracy of the classification models.This research addresses the challenge of enhancing the image retrieval system by developing a novel approach,EfficientNet-Convolutional Neural Network(EffNet-CNN).The key objective of this research is to evaluate the proposed EffNet-CNN model’s performance in image classification,image mining,and CBIR.The novelty of the proposed EffNet-CNN model includes the integration of different techniques and modifications.The model includes the Mahalanobis distance metric for feature matching,which enhances the similarity measurements.The model extends EfficientNet architecture by incorporating additional convolutional layers,batch normalization,dropout,and pooling layers for improved hierarchical feature extraction.A systematic hyperparameter optimization using SGD,performance evaluation with three datasets,and data normalization for improving feature representations.The EffNet-CNN is assessed utilizing precision,accuracy,F-measure,and recall metrics across MS-COCO,CIFAR-10 and 100 datasets.The model achieved accuracy values ranging from 90.60%to 95.90%for the MS-COCO dataset,96.8%to 98.3%for the CIFAR-10 dataset and 92.9%to 98.6%for the CIFAR-100 dataset.A validation of the EffNet-CNN model’s results with other models reveals the proposed model’s superior performance.The results highlight the potential of the EffNet-CNN model proposed for image classification and its usefulness in image mining and CBIR.展开更多
Retrieval analysis in total knee arthroplasty(TKA)has been little studied in the literature.A narrative review of the literature to understand the current importance of retrieval analysis in TKA has been conducted.On ...Retrieval analysis in total knee arthroplasty(TKA)has been little studied in the literature.A narrative review of the literature to understand the current importance of retrieval analysis in TKA has been conducted.On August 27,2024,a literature search was performed in PubMed using“TKA retrieval analysis”as keywords.A total of 160 articles were found,of which only 19 were analyzed because they were directly related to the subject of this article.Rotating-platform(mobile-bearing)TKA has no surface damage advantage over fixed-bearing TKA.TKAs with central locking mechanisms are more prone to debond from the cement mantle.No major wear of the polyethylene(PE)component in TKA using oxidized zirconium components occurs.Femoral components of cobalt-chromium roughen more than oxidized zirconium femoral components.The use of a polished tibial tray over an unpolished design is advised.At short-run assessment(15 months on average),antioxidant-stabilized highly crosslinked PE components are not clinically different in surface damage,density of crosslinking,or oxidation compared to standard remelted highly crosslinked PE components.A correlation between implant position and PE component surface damage has been reported.It shows the importance of optimizing component position to reduce PE component damage.Contemporary knee tumor megaendoprostheses show notable volumetric metal wear originated at the rotating hinge.Retrieval analysis in TKA renders relevant data on how different prosthetic designs described in the literature perform.Such information can help to improve future prosthetic designs to increase prosthetic survival.展开更多
Ciphertext data retrieval in cloud databases suffers from some critical limitations,such as inadequate security measures,disorganized key management practices,and insufficient retrieval access control capabilities.To ...Ciphertext data retrieval in cloud databases suffers from some critical limitations,such as inadequate security measures,disorganized key management practices,and insufficient retrieval access control capabilities.To address these problems,this paper proposes an enhanced Fully Homomorphic Encryption(FHE)algorithm based on an improved DGHV algorithm,coupled with an optimized ciphertext retrieval scheme.Our specific contributions are outlined as follows:First,we employ an authorization code to verify the user’s retrieval authority and perform hierarchical access control on cloud storage data.Second,a triple-key encryption mechanism,which separates the data encryption key,retrieval authorization key,and retrieval key,is designed.Different keys are provided to different entities to run corresponding system functions.The key separation architecture proves particularly advantageous in multi-verifier coexistence scenarios,environments involving untrusted third-party retrieval services.Finally,the enhanced DGHV-based retrieval mechanism extends conventional functionality by enabling multi-keyword queries with similarity-ranked results,thereby significantly improving both the functionality and usability of the FHE system.展开更多
Recent advancements in large language models(LLMs)have driven remarkable progress in text process-ing,opening new avenues for medical knowledge discovery.In this study,we present ERQA,a mEdical knowledge Retrieval and...Recent advancements in large language models(LLMs)have driven remarkable progress in text process-ing,opening new avenues for medical knowledge discovery.In this study,we present ERQA,a mEdical knowledge Retrieval and Question-Answering framework powered by an enhanced LLM that integrates a semantic vector database and a curated literature repository.The ERQA framework leverages domain-specific incremental pretraining and conducts supervised fine-tuning on medical literature,enabling retrieval and question-answering(QA)tasks to be completed with high precision.Performance evaluations implemented on the coronavirus disease 2019(COVID-19)and TripClick data-sets demonstrate the robust capabilities of ERQA across multiple tasks.On the COVID-19 dataset,ERQA-13B achieves state-of-the-art retrieval metrics,with normalized discounted cumulative gain at top 10(NDCG@10)0.297,recall values at top 10(Recall@10)0.347,and mean reciprocal rank(MRR)=0.370;it also attains strong abstract summarization performance,with a recall-oriented understudy for gisting evaluation(ROUGE)-1 score of 0.434,and QA performance,with a bilingual evaluation understudy(BLEU)-1 score of 7.851.The comparable performance achieved on the TripClick dataset further under-scores the adaptability of ERQA across diverse medical topics.These findings suggest that ERQA repre-sents a significant step toward efficient biomedical knowledge retrieval and QA.展开更多
Accurate retrieval of casting 3D models is crucial for process reuse.Current methods primarily focus on shape similarity,neglecting process design features,which compromises reusability.In this study,a novel deep lear...Accurate retrieval of casting 3D models is crucial for process reuse.Current methods primarily focus on shape similarity,neglecting process design features,which compromises reusability.In this study,a novel deep learning retrieval method for process reuse was proposed,which integrates process design features into the retrieval of casting 3D models.This method leverages the comparative language-image pretraining(CLIP)model to extract shape features from the three views and sectional views of the casting model and combines them with process design features such as modulus,main wall thickness,symmetry,and length-to-height ratio to enhance process reusability.A database of 230 production casting models was established for model validation.Results indicate that incorporating process design features improves model accuracy by 6.09%,reaching 97.82%,and increases process similarity by 30.25%.The reusability of the process was further verified using the casting simulation software EasyCast.The results show that the process retrieved after integrating process design features produces the least shrinkage in the target model,demonstrating this method’s superior ability for process reuse.This approach does not require a large dataset for training and optimization,making it highly applicable to casting process design and related manufacturing processes.展开更多
Video imagery enables both qualitative characterization and quantitative retrieval of low-visibility conditions.These phenomena exhibit complex nonlinear dependencies on atmospheric processes,particularly during moist...Video imagery enables both qualitative characterization and quantitative retrieval of low-visibility conditions.These phenomena exhibit complex nonlinear dependencies on atmospheric processes,particularly during moisture-driven weather events such as fog,rain,and snow.To address this challenge,we propose a dual-branch neural architecture that synergistically processes optical imagery and multi-source meteorological data(temperature,humidity,and wind speed).The framework employs a convolutional neural network(CNN)branch to extract visibility-related visual features from video imagery sequences,while a parallel artificial neural network(ANN)branch decodes nonlinear relationships among the meteorological factors.Cross-modal feature fusion is achieved through an adaptive weighting layer.To validate the framework,multimodal Backpropagation-VGG(BP-VGG)and Backpropagation-ResNet(BP-ResNet)models are developed and trained/tested using historical imagery and meteorological observations from Nanjing Lukou International Airport.The results demonstrate that the multimodal networks reduce retrieval errors by approximately 8%–10%compared to unimodal networks relying solely on imagery.Among the multimodal models,BP-ResNet exhibits the best performance with a mean absolute percentage error(MAPE)of 8.5%.Analysis of typical case studies reveals that visibility fluctuates rapidly while meteorological factors change gradually,highlighting the crucial role of high-frequency imaging data in intelligent visibility retrieval models.The superior performance of BP-ResNet over BP-VGG is attributed to its use of residual blocks,which enables BP-ResNet to excel in multimodal processing by effectively leveraging data complementarity for synergistic improvements.This study presents an end-to-end intelligent visibility inversion framework that directly retrieves visibility values,enhancing its applicability across industries.However,while this approach boosts accuracy and applicability,its performance in critical low-visibility scenarios remains suboptimal,necessitating further research into more advanced retrieval techniques—particularly under extreme visibility conditions.展开更多
Aerosol optical depth(AOD)and fine particulate matter with a diameter of less than or equal to 2.5μm(PM_(2.5))play crucial roles in air quality,human health,and climate change.However,the complex correlation of AOD–...Aerosol optical depth(AOD)and fine particulate matter with a diameter of less than or equal to 2.5μm(PM_(2.5))play crucial roles in air quality,human health,and climate change.However,the complex correlation of AOD–PM_(2.5)and the limitations of existing algorithms pose a significant challenge in realizing the accurate joint retrieval of these two parameters at the same location.On this point,a multi-task learning(MTL)model,which enables the joint retrieval of PM_(2.5)concentration and AOD,is proposed and applied on the top-of-the-atmosphere reflectance data gathered by the Fengyun-4A Advanced Geosynchronous Radiation Imager(FY-4A AGRI),and compared to that of two single-task learning models—namely,Random Forest(RF)and Deep Neural Network(DNN).Specifically,MTL achieves a coefficient of determination(R^(2))of 0.88 and a root-mean-square error(RMSE)of 0.10 in AOD retrieval.In comparison to RF,the R^(2)increases by 0.04,the RMSE decreases by 0.02,and the percentage of retrieval results falling within the expected error range(Within-EE)rises by 5.55%.The R^(2)and RMSE of PM_(2.5)retrieval by MTL are 0.84 and 13.76μg m~(-3)respectively.Compared with RF,the R^(2)increases by 0.06,the RMSE decreases by 4.55μg m~(-3),and the Within-EE increases by 7.28%.Additionally,compared to DNN,MTL shows an increase of 0.01 in R^(2)and a decrease of 0.02 in RMSE in AOD retrieval,with a corresponding increase of 2.89%in Within-EE.For PM_(2.5)retrieval,MTL exhibits an increase of 0.05 in R^(2),a decrease of 1.76μg m~(-3)in RMSE,and an increase of 6.83%in Within-EE.The evaluation suggests that MTL is able to provide simultaneously improved AOD and PM_(2.5)retrievals,demonstrating a significant advantage in efficiently capturing the spatial distribution of PM_(2.5)concentration and AOD.展开更多
Numerous studies on the formation and consolidation of memory have shown that memory processes are characterized by phase-dependent and dynamic regulation.Memory retrieval,as the only representation of memory content ...Numerous studies on the formation and consolidation of memory have shown that memory processes are characterized by phase-dependent and dynamic regulation.Memory retrieval,as the only representation of memory content and an active form of memory processing that induces memory reconsolidation,has attracted increasing attention in recent years.Although the molecular mechanisms specifc to memory retrievalinduced reconsolidation have been gradually revealed,an understanding of the time-dependent regulatory mechanisms of this process is still lacking.In this study,we applied a transcriptome analysis of memory retrieval at diferent time points in the recent memory stage.Diferential expression analysis and Short Time-series Expression Miner(STEM)depicting temporal gene expression patterns indicated that most diferential gene expression occurred at 48 h,and the STEM cluster showing the greatest transcriptional upregulation at 48 h demonstrated the most significant diference.We then screened the diferentially-expressed genes associated with that met the expression patterns of those cluster-identifed genes that have been reported to be involved in learning and memory processes in addition to dipeptidyl peptidase 9(DPP9).Further quantitative polymerase chain reaction verifcation and pharmacological intervention suggested that DPP9 is involved in 48-h fear memory retrieval and viral vector-mediated overexpression of DPP9 countered the 48-h retrieval-induced attenuation of fear memory.Taken together,our fndings suggest that temporal gene expression patterns are induced by recent memory retrieval and provide hitherto undocumented evidence of the role of DPP9 in the retrieval-induced reconsolidation of fear memory.展开更多
基金supported by the National Key R&D Program of China(No.2022ZD0118402)。
文摘Remote sensing cross-modal image-text retrieval(RSCIR)can flexibly and subjectively retrieve remote sensing images utilizing query text,which has received more researchers’attention recently.However,with the increasing volume of visual-language pre-training model parameters,direct transfer learning consumes a substantial amount of computational and storage resources.Moreover,recently proposed parameter-efficient transfer learning methods mainly focus on the reconstruction of channel features,ignoring the spatial features which are vital for modeling key entity relationships.To address these issues,we design an efficient transfer learning framework for RSCIR,which is based on spatial feature efficient reconstruction(SPER).A concise and efficient spatial adapter is introduced to enhance the extraction of spatial relationships.The spatial adapter is able to spatially reconstruct the features in the backbone with few parameters while incorporating the prior information from the channel dimension.We conduct quantitative and qualitative experiments on two different commonly used RSCIR datasets.Compared with traditional methods,our approach achieves an improvement of 3%-11% in sumR metric.Compared with methods finetuning all parameters,our proposed method only trains less than 1% of the parameters,while maintaining an overall performance of about 96%.
基金sponsored by Natural Science Foundation of Xinjiang Uygur Autonomous Region(2024D01A19).
文摘Research on mass gathering events is critical for ensuring public security and maintaining social order.However,most of the existing works focus on crowd behavior analysis areas such as anomaly detection and crowd counting,and there is a relative lack of research on mass gathering behaviors.We believe real-time detection and monitoring of mass gathering behaviors are essential formigrating potential security risks and emergencies.Therefore,it is imperative to develop a method capable of accurately identifying and localizing mass gatherings before disasters occur,enabling prompt and effective responses.To address this problem,we propose an innovative Event-Driven Attention Network(EDAN),which achieves image-text matching in the scenario of mass gathering events with good results for the first time.Traditional image-text retrieval methods based on global alignment are difficult to capture the local details within complex scenes,limiting retrieval accuracy.While local alignment-based methods aremore effective at extracting detailed features,they frequently process raw textual features directly,which often contain ambiguities and redundant information that can diminish retrieval efficiency and degrade model performance.To overcome these challenges,EDAN introduces an Event-Driven AttentionModule that adaptively focuses attention on image regions or textual words relevant to the event type.By calculating the semantic distance between event labels and textual content,this module effectively significantly reduces computational complexity and enhances retrieval efficiency.To validate the effectiveness of EDAN,we construct a dedicated multimodal dataset tailored for the analysis of mass gathering events,providing a reliable foundation for subsequent studies.We conduct comparative experiments with other methods on our dataset,the experimental results demonstrate the effectiveness of EDAN.In the image-to-text retrieval task,EDAN achieved the best performance on the R@5 metric,while in the text-to-image retrieval task,it showed superior results on both R@10 and R@5 metrics.Additionally,EDAN excelled in the overall Rsummetric,achieving the best performance.Finally,ablation studies further verified the effectiveness of event-driven attention module.
文摘With the incessant propulsion of the Open Door Policy,which is related to the consolidation of international collaborative partnerships,an increasing number of Chinese companies are moving toward cooperating countries to participate in infrastructure construction,employing a win-win strategy in favor of the people and governments of both countries.Among the cooperation domains,our country’s electrical companies have achieved a series of remarkable results in the international Engineering,Procurement,and Construction(EPC)project market with their outstanding business capabilities and technical advantages.Nevertheless,some shortcomings cannot be overlooked,the most notable of which appears to be the impediment associated with engineering translation,which has always been an obsession among translators of Chinese companies.Taking the transmission line project in the Republic of Madagascar as an example,an analysis of French-Chinese translation methods of electrical engineering terminology in the field of the transmission line is carried out.
基金This work was partially supported by Chongqing Natural Science Foundation of China(Grant No.CSTB2022NSCQ-MSX1417)the Science and Technology Research Program of Chongqing Municipal Education Commission(Grant No.KJZD-K202200513)+2 种基金Chongqing Normal University Fund(Grant No.22XLB003)Chongqing Education Science Planning Project(Grant No.2021-GX-320)Humanities and Social Sciences Project of Chongqing Education Commission of China(Grant No.22SKGH100).
文摘In recent years,cross-modal hash retrieval has become a popular research field because of its advantages of high efficiency and low storage.Cross-modal retrieval technology can be applied to search engines,crossmodalmedical processing,etc.The existing main method is to use amulti-label matching paradigm to finish the retrieval tasks.However,such methods do not use fine-grained information in the multi-modal data,which may lead to suboptimal results.To avoid cross-modal matching turning into label matching,this paper proposes an end-to-end fine-grained cross-modal hash retrieval method,which can focus more on the fine-grained semantic information of multi-modal data.First,the method refines the image features and no longer uses multiple labels to represent text features but uses BERT for processing.Second,this method uses the inference capabilities of the transformer encoder to generate global fine-grained features.Finally,in order to better judge the effect of the fine-grained model,this paper uses the datasets in the image text matching field instead of the traditional label-matching datasets.This article experiment on Microsoft COCO(MS-COCO)and Flickr30K datasets and compare it with the previous classicalmethods.The experimental results show that this method can obtain more advanced results in the cross-modal hash retrieval field.
基金supported by the startup fund from Beijing Normal University(312200502560 to J.Z.)the National Natural Science Foundation of China(32501383 to J.Z.,and 31672297 and 32271559 to W.D.)the Max Planck Society(to B.K.)。
文摘The arms race between avian brood parasites and their hosts provides a classic model for studying coevolution.In one of the most widespread obligate brood parasites,the Common Cuckoo(Cuculus canorus),chicks typically evict all host progeny(eggs and nestlings)from the nest cup,resulting in complete reproductive failure for the host.Host parents of Common Cuckoos could thus potentially benefit from retrieving evicted eggs and nestlings into the nest cup.However,whether hosts of the Common Cuckoo exhibit such retrieval behavior has been scarcely studied.In this study,we experimentally investigated the occurrence of retrieval in a nestbox-breeding population of Daurian Redstarts(Phoenicurus auroreus),a common cavity-nesting host of the Common Cuckoo.To test the redstarts'response to an egg or a nestling outside the nest cup,we experimentally placed either a conspecific egg,a model cuckoo egg,or a redstart nestling near the rim of the nest cup.We found that redstarts never showed retrieval behavior of either eggs or nestlings.All hosts ignored the experimental nestling and conspecific egg,but most ejected the model cuckoo egg from the nestbox.Our results suggest that selection for retrieval behavior in this cavity-nesting host may be weak or even negative.We discuss several ecological and evolutionary factors that may explain the absence of retrieval in this system.
基金supported by the National Natural Science Foundation of China under Grant U2442219Fengyun Satellite Application Pioneer Program(2023)Special Initiative on Numerical Weather Prediction(NWP)Applications,the Civil Aerospace Technology Pre-Research Project(D040405)the Joint Funds of the Zhejiang Provincial Natural Science Foundation of China under Grant No.LZJMZ23D050003。
文摘Accurate retrieval of atmospheric vertical profiles is critical for improving weather prediction and climate monitoring.However,the complexity of atmospheric processes in cloudy regions poses challenges compared to those of clear sky scenarios.This study presents a novel framework that integrates Bayesian optimization and machine learning approaches to retrieve atmospheric vertical profiles—including temperature,humidity,ozone concentration,cloud fraction,ice water content(IWC),and liquid water content(LWC)—from hyperspectral infrared observations.Specifically,a Bayesian method was used to refine ERA5 reanalysis data by minimizing brightness temperature(BT)discrepancies against FY-4B Geostationary Interferometric Infrared Sounder(GIIRS)observations,generating a high-quality profile database(~2.8 million profiles)across diverse weather systems.The optimized profiles improve radiative consistency,reducing BT biases from>40 K to<10 K in cloudy regions.To further overcome the limitations of the Bayesian method,we developed a Transformer-Resnet hybrid model(TERNet),which achieved superior performance with RMSE values of 1.61 K(temperature),5.77%(humidity),and 2.25×10^(–6)/6.09×10^(–6)kg kg^(–1)(IWC/LWC)across the entire vertical levels in all-sky conditions.The TERNet outperforms both ERA5 in cloud parameter retrieval and the GIIRS L2 product in thermodynamic profiling.Independent verification with radiosonde and Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations(CALIPSO)datasets confirms the framework's reliability across various meteorological regimes.This work demonstrates the capability of combining physics-informed Bayesian methods with data-driven machine learning to fully exploit hyperspectral IR data.
文摘Background:Hormonal treatment and response as a predictor of sperm retrieval prior to microdissection testicular sperm extraction(micro-TESE)are not well established in the current literature.This study aimed to investigate the hormonal response as a predictor of sperm retrieval among men with nonobstructive azoospermia(NOA).Methods:Seventy-seven consecutive patients who had testosterone levels≤14 nmol/L were treated medically with an aromatase inhibitor or recombinant human chorionic gonadotropin(rec-hCG)prior to micro-TESE and were included.Thirty-four(44.2%)had unexplained NOA(UNEX),25(32.5%)had Klinefelter syndrome(KS),8(10.4%)had a history of cryptorchidism(UDT),4(5.2%)had microdeletion of the Azoospermia factor C(AZFc),and 6(7.8%)were treated previously with chemotherapy.Baseline and post-treatment serum hormonal levels were documented.Pre-op testosterone levels were entered into binary logistic regressions with age,Follicle-stimulating hormone(FSH),and Luteinizing hormone(LH)levels to test for significance with sperm retrieval.We then built logistic regression models to identify predictors of successful surgical sperm retrieval(SSR).Results:Forty-five patients(58%)had successful retrieval.In 32 patients(42%),no sperm was retrieved.Both the mean pre-op testosterone and the mean testosterone change between the two groups were significant(p=0.02 and p=0.011,respectively).Receiver operating characteristic(ROC)analysis demonstrated an area under the curve(AUC)of 0.785(95%CI=0.685-0.886,p<0.001).The Youden index coefficient was calculated for KS and UNEX.The cut-off point for KS was established at 0.764(sensitivity=0.875,false positive rate[FPR]=0.111),and 0.215 for UNEX(sensitivity=0.438,FPR=0.222).We also observed a correlation between age and SSR(p=0.05).In KS patients,SSR was determined by pre-op testosterone levels irrespective of age.Conclusion:Pre-operative hormonal response is a predictor for SSR in NOA patients who were treated medically.This data may help during pre-operative counselling.
文摘Cross-modal image-text retrieval is a fundamental task in bridging vision and language. It faces two main challenges that are typically not well addressed in previous works. 1) Generalizability: Existing methods often assume a strong semantic correlation between each text-image pair, which are thus difficult to generalize to real-world scenarios where the weak correlation dominates. 2) Efficiency: Many latest works adopt the single-tower architecture with heavy detectors, which are inefficient during the inference stage because the costly computation needs to be repeated for each text-image pair. In this work, to overcome these two challenges, we propose a two-tower cross-modal contrastive learning (CMCL) framework. Specifically, we first devise a two-tower architecture, which enables a unified feature space for the text and image modalities to be directly compared with each other, alleviating the heavy computation during inference. We further introduce a simple yet effective module named multi-grid split (MGS) to learn fine-grained image features without using detectors. Last but not the least, we deploy a cross-modal contrastive loss on the global image/text features to learn their weak correlation and thus achieve high generalizability. To validate that our CMCL can be readily generalized to real-world scenarios, we construct a large multi-source image-text dataset called weak semantic correlation dataset (WSCD). Extensive experiments show that our CMCL outperforms the state-of-the-arts while being much more efficient.
基金supported by the National Natural Science Foundation of China under Grant No.62076048.
文摘Image-text retrieval aims to capture the semantic correspondence between images and texts,which serves as a foundation and crucial component in multi-modal recommendations,search systems,and online shopping.Existing mainstream methods primarily focus on modeling the association of image-text pairs while neglecting the advantageous impact of multi-task learning on image-text retrieval.To this end,a multi-task visual semantic embedding network(MVSEN)is proposed for image-text retrieval.Specifically,we design two auxiliary tasks,including text-text matching and multi-label classification,for semantic constraints to improve the generalization and robustness of visual semantic embedding from a training perspective.Besides,we present an intra-and inter-modality interaction scheme to learn discriminative visual and textual feature representations by facilitating information flow within and between modalities.Subsequently,we utilize multi-layer graph convolutional networks in a cascading manner to infer the correlation of image-text pairs.Experimental results show that MVSEN outperforms state-of-the-art methods on two publicly available datasets,Flickr30K and MSCOCO,with rSum improvements of 8.2%and 3.0%,respectively.
基金funded by the Institute of InformationTechnology,VietnamAcademy of Science and Technology(project number CSCL02.02/22-23)“Research and Development of Methods for Searching Similar Trademark Images Using Machine Learning to Support Trademark Examination in Vietnam”.
文摘Image-based similar trademark retrieval is a time-consuming and labor-intensive task in the trademark examination process.This paper aims to support trademark examiners by training Deep Convolutional Neural Network(DCNN)models for effective Trademark Image Retrieval(TIR).To achieve this goal,we first develop a novel labeling method that automatically generates hundreds of thousands of labeled similar and dissimilar trademark image pairs using accompanying data fields such as citation lists,Vienna classification(VC)codes,and trademark ownership information.This approach eliminates the need for manual labeling and provides a large-scale dataset suitable for training deep learning models.We then train DCNN models based on Siamese and Triplet architectures,evaluating various feature extractors to determine the most effective configuration.Furthermore,we present an Adapted Contrastive Loss Function(ACLF)for the trademark retrieval task,specifically engineered to mitigate the influence of noisy labels found in automatically created datasets.Experimental results indicate that our proposed model(Efficient-Net_v21_Siamese)performs best at both True Negative Rate(TNR)threshold levels,TNR 0.9 and TNR 0.95,with==respective True Positive Rates(TPRs)of 77.7%and 70.8%and accuracies of 83.9%and 80.4%.Additionally,when testing on the public trademark dataset METU_v2,our model achieves a normalized average rank(NAR)of 0.0169,outperforming the current state-of-the-art(SOTA)model.Based on these findings,we estimate that considering only approximately 10%of the returned trademarks would be sufficient,significantly reducing the review time.Therefore,the paper highlights the potential of utilizing national trademark data to enhance the accuracy and efficiency of trademark retrieval systems,ultimately supporting trademark examiners in their evaluation tasks.
基金supported by the National Science Foundations of China(No.61905256)the National Key Research and Development Program of China(No.2019YFC0214702)the Youth Innovation Promotion Association of Chinese Academy of Sciences(No.2020439)。
文摘This study describes the use of the weighted multiplicative algebraic reconstruction technique(WMART)to obtain vertical ozone profiles from limb observations performed by the scanning imaging absorption spectrometer for atmospheric chartography(SCIAMACHY).This technique is based on SaskMART(the combination of the multiplicative algebraic reconstruction technique and SaskTRAN radiative transfer model),which was originally developed for optical spectrometer and infrared imaging system(OSIRIS)data.One of the objectives of this study was to obtain consistent ozone profiles from the two satellites.In this study,the WMART algorithm is combined with a radiative transfer model(SCIATRAN),as well as a set of measurement vectors comprising five Hartley pairing vectors(HPVs)and one Chappuis triplet vector(CTV),to retrieve ozone profiles in the altitude range of 10–69 km.Considering that the weighting factors in WMART have a significant effect on the retrievals,we propose a novel approach to calculate the pair/triplet weighting factors using wavelength weighting functions.The results of the application of the proposed ozone retrieval scheme are compared with the SCIAMACHY v3.5 ozone product by University of Bremen and validated against profiles derived from other passive satellite observations or measured by ozonesondes.Between 18 and 55 km,the retrieved ozone profiles typically agree with data from the SCIAMACHY ozone product within 5%for tropics and middle latitudes,whereas a negative deviation exists between 35 and 50 km for northern high latitudes,with a deviation of less than 10%above 50 km.Comparison of the retrieved profiles with microwave limb sounder(MLS)v5.0 indicates that the difference is within±5%between 18 and 55 km,and an agreement within 10%is achieved in other altitudes for tropics and middle latitudes.Comparison of the retrieved profiles with OSIRIS v7.1 indicates that the average deviation is within±5%between 20 and 59 km,and difference of approximately 10%is achieved below 20 km.Compared with ozonesondes data,a general validity of the retrievals is no more than 5%between 15 and 30 km.
基金supported by the National Natural Science Foundation of China(T2394531)the National Key R&D Program of China(2024YFF1206500)+1 种基金the Shanghai Municipal Science and Technology Major Project(2018SHZDZX01)ZJ Lab,and the Shanghai Center for Brain Science and Brain-Inspired Technology,China.
文摘DDeeaarr EEddiittoorr,,The encoding and retrieval of emotional memories demands intricate interplay within the limbic network,where the network state is subject to significant reconfiguration by learning-induced plasticity,behavioral state,and contextual information[1].
基金The authors extend their appreciation to the Deanship of Research and Graduate Studies at King Khalid University,Kingdom of Saudi Arabia,for funding this work through the Small Research Group Project under Grant Number RGP.1/316/45.
文摘Content-Based Image Retrieval(CBIR)and image mining are becoming more important study fields in computer vision due to their wide range of applications in healthcare,security,and various domains.The image retrieval system mainly relies on the efficiency and accuracy of the classification models.This research addresses the challenge of enhancing the image retrieval system by developing a novel approach,EfficientNet-Convolutional Neural Network(EffNet-CNN).The key objective of this research is to evaluate the proposed EffNet-CNN model’s performance in image classification,image mining,and CBIR.The novelty of the proposed EffNet-CNN model includes the integration of different techniques and modifications.The model includes the Mahalanobis distance metric for feature matching,which enhances the similarity measurements.The model extends EfficientNet architecture by incorporating additional convolutional layers,batch normalization,dropout,and pooling layers for improved hierarchical feature extraction.A systematic hyperparameter optimization using SGD,performance evaluation with three datasets,and data normalization for improving feature representations.The EffNet-CNN is assessed utilizing precision,accuracy,F-measure,and recall metrics across MS-COCO,CIFAR-10 and 100 datasets.The model achieved accuracy values ranging from 90.60%to 95.90%for the MS-COCO dataset,96.8%to 98.3%for the CIFAR-10 dataset and 92.9%to 98.6%for the CIFAR-100 dataset.A validation of the EffNet-CNN model’s results with other models reveals the proposed model’s superior performance.The results highlight the potential of the EffNet-CNN model proposed for image classification and its usefulness in image mining and CBIR.
文摘Retrieval analysis in total knee arthroplasty(TKA)has been little studied in the literature.A narrative review of the literature to understand the current importance of retrieval analysis in TKA has been conducted.On August 27,2024,a literature search was performed in PubMed using“TKA retrieval analysis”as keywords.A total of 160 articles were found,of which only 19 were analyzed because they were directly related to the subject of this article.Rotating-platform(mobile-bearing)TKA has no surface damage advantage over fixed-bearing TKA.TKAs with central locking mechanisms are more prone to debond from the cement mantle.No major wear of the polyethylene(PE)component in TKA using oxidized zirconium components occurs.Femoral components of cobalt-chromium roughen more than oxidized zirconium femoral components.The use of a polished tibial tray over an unpolished design is advised.At short-run assessment(15 months on average),antioxidant-stabilized highly crosslinked PE components are not clinically different in surface damage,density of crosslinking,or oxidation compared to standard remelted highly crosslinked PE components.A correlation between implant position and PE component surface damage has been reported.It shows the importance of optimizing component position to reduce PE component damage.Contemporary knee tumor megaendoprostheses show notable volumetric metal wear originated at the rotating hinge.Retrieval analysis in TKA renders relevant data on how different prosthetic designs described in the literature perform.Such information can help to improve future prosthetic designs to increase prosthetic survival.
基金supported by the Innovation Program for Quantum Science and technology(2021ZD0301300)supported by the Fundamental Research Funds for the Central Universities(Nos.3282024046,3282024052,3282024058,3282023017).
文摘Ciphertext data retrieval in cloud databases suffers from some critical limitations,such as inadequate security measures,disorganized key management practices,and insufficient retrieval access control capabilities.To address these problems,this paper proposes an enhanced Fully Homomorphic Encryption(FHE)algorithm based on an improved DGHV algorithm,coupled with an optimized ciphertext retrieval scheme.Our specific contributions are outlined as follows:First,we employ an authorization code to verify the user’s retrieval authority and perform hierarchical access control on cloud storage data.Second,a triple-key encryption mechanism,which separates the data encryption key,retrieval authorization key,and retrieval key,is designed.Different keys are provided to different entities to run corresponding system functions.The key separation architecture proves particularly advantageous in multi-verifier coexistence scenarios,environments involving untrusted third-party retrieval services.Finally,the enhanced DGHV-based retrieval mechanism extends conventional functionality by enabling multi-keyword queries with similarity-ranked results,thereby significantly improving both the functionality and usability of the FHE system.
基金supported by the Innovation Fund for Medical Sciences of the Chinese Academy of Medical Sciences(2021-I2M-1-033)the National Key Research and Development Program of China(2022YFF0711900).
文摘Recent advancements in large language models(LLMs)have driven remarkable progress in text process-ing,opening new avenues for medical knowledge discovery.In this study,we present ERQA,a mEdical knowledge Retrieval and Question-Answering framework powered by an enhanced LLM that integrates a semantic vector database and a curated literature repository.The ERQA framework leverages domain-specific incremental pretraining and conducts supervised fine-tuning on medical literature,enabling retrieval and question-answering(QA)tasks to be completed with high precision.Performance evaluations implemented on the coronavirus disease 2019(COVID-19)and TripClick data-sets demonstrate the robust capabilities of ERQA across multiple tasks.On the COVID-19 dataset,ERQA-13B achieves state-of-the-art retrieval metrics,with normalized discounted cumulative gain at top 10(NDCG@10)0.297,recall values at top 10(Recall@10)0.347,and mean reciprocal rank(MRR)=0.370;it also attains strong abstract summarization performance,with a recall-oriented understudy for gisting evaluation(ROUGE)-1 score of 0.434,and QA performance,with a bilingual evaluation understudy(BLEU)-1 score of 7.851.The comparable performance achieved on the TripClick dataset further under-scores the adaptability of ERQA across diverse medical topics.These findings suggest that ERQA repre-sents a significant step toward efficient biomedical knowledge retrieval and QA.
基金supported by the National Natural Science Foundation of China(Nos.52074246,52275390,52375394)the National Defense Basic Scientific Research Program of China(No.JCKY2020408B002)the Key R&D Program of Shanxi Province(No.202102050201011).
文摘Accurate retrieval of casting 3D models is crucial for process reuse.Current methods primarily focus on shape similarity,neglecting process design features,which compromises reusability.In this study,a novel deep learning retrieval method for process reuse was proposed,which integrates process design features into the retrieval of casting 3D models.This method leverages the comparative language-image pretraining(CLIP)model to extract shape features from the three views and sectional views of the casting model and combines them with process design features such as modulus,main wall thickness,symmetry,and length-to-height ratio to enhance process reusability.A database of 230 production casting models was established for model validation.Results indicate that incorporating process design features improves model accuracy by 6.09%,reaching 97.82%,and increases process similarity by 30.25%.The reusability of the process was further verified using the casting simulation software EasyCast.The results show that the process retrieved after integrating process design features produces the least shrinkage in the target model,demonstrating this method’s superior ability for process reuse.This approach does not require a large dataset for training and optimization,making it highly applicable to casting process design and related manufacturing processes.
基金Foundation of Key Laboratory of Smart Earth(KF2023ZD03-02)China Meteorological Administration Innovation development project(CXFZ2025J116)+1 种基金National Natural Science Foundation of China(42205197)Basic Research Fund of CAMS(2022Y023,2022Y025)。
文摘Video imagery enables both qualitative characterization and quantitative retrieval of low-visibility conditions.These phenomena exhibit complex nonlinear dependencies on atmospheric processes,particularly during moisture-driven weather events such as fog,rain,and snow.To address this challenge,we propose a dual-branch neural architecture that synergistically processes optical imagery and multi-source meteorological data(temperature,humidity,and wind speed).The framework employs a convolutional neural network(CNN)branch to extract visibility-related visual features from video imagery sequences,while a parallel artificial neural network(ANN)branch decodes nonlinear relationships among the meteorological factors.Cross-modal feature fusion is achieved through an adaptive weighting layer.To validate the framework,multimodal Backpropagation-VGG(BP-VGG)and Backpropagation-ResNet(BP-ResNet)models are developed and trained/tested using historical imagery and meteorological observations from Nanjing Lukou International Airport.The results demonstrate that the multimodal networks reduce retrieval errors by approximately 8%–10%compared to unimodal networks relying solely on imagery.Among the multimodal models,BP-ResNet exhibits the best performance with a mean absolute percentage error(MAPE)of 8.5%.Analysis of typical case studies reveals that visibility fluctuates rapidly while meteorological factors change gradually,highlighting the crucial role of high-frequency imaging data in intelligent visibility retrieval models.The superior performance of BP-ResNet over BP-VGG is attributed to its use of residual blocks,which enables BP-ResNet to excel in multimodal processing by effectively leveraging data complementarity for synergistic improvements.This study presents an end-to-end intelligent visibility inversion framework that directly retrieves visibility values,enhancing its applicability across industries.However,while this approach boosts accuracy and applicability,its performance in critical low-visibility scenarios remains suboptimal,necessitating further research into more advanced retrieval techniques—particularly under extreme visibility conditions.
基金supported by the National Natural Science Foundation of China(Grant Nos.42030708,42375138,42030608,42105128,42075079)the Opening Foundation of Key Laboratory of Atmospheric Sounding,China Meteorological Administration(CMA),and the CMA Research Center on Meteorological Observation Engineering Technology(Grant No.U2021Z03),and the Opening Foundation of the Key Laboratory of Atmospheric Chemistry,CMA(Grant No.2022B02)。
文摘Aerosol optical depth(AOD)and fine particulate matter with a diameter of less than or equal to 2.5μm(PM_(2.5))play crucial roles in air quality,human health,and climate change.However,the complex correlation of AOD–PM_(2.5)and the limitations of existing algorithms pose a significant challenge in realizing the accurate joint retrieval of these two parameters at the same location.On this point,a multi-task learning(MTL)model,which enables the joint retrieval of PM_(2.5)concentration and AOD,is proposed and applied on the top-of-the-atmosphere reflectance data gathered by the Fengyun-4A Advanced Geosynchronous Radiation Imager(FY-4A AGRI),and compared to that of two single-task learning models—namely,Random Forest(RF)and Deep Neural Network(DNN).Specifically,MTL achieves a coefficient of determination(R^(2))of 0.88 and a root-mean-square error(RMSE)of 0.10 in AOD retrieval.In comparison to RF,the R^(2)increases by 0.04,the RMSE decreases by 0.02,and the percentage of retrieval results falling within the expected error range(Within-EE)rises by 5.55%.The R^(2)and RMSE of PM_(2.5)retrieval by MTL are 0.84 and 13.76μg m~(-3)respectively.Compared with RF,the R^(2)increases by 0.06,the RMSE decreases by 4.55μg m~(-3),and the Within-EE increases by 7.28%.Additionally,compared to DNN,MTL shows an increase of 0.01 in R^(2)and a decrease of 0.02 in RMSE in AOD retrieval,with a corresponding increase of 2.89%in Within-EE.For PM_(2.5)retrieval,MTL exhibits an increase of 0.05 in R^(2),a decrease of 1.76μg m~(-3)in RMSE,and an increase of 6.83%in Within-EE.The evaluation suggests that MTL is able to provide simultaneously improved AOD and PM_(2.5)retrievals,demonstrating a significant advantage in efficiently capturing the spatial distribution of PM_(2.5)concentration and AOD.
基金supported by the STI2030-Major Projects(2022ZD0204900)the National Natural Science Foundation of China(32071029 and 32271080)+1 种基金the Strategic Priority Research Program of the Chinese Academy of Sciences(XDB32020200)the Yunnan Provincial Science and Technology Department(202402AA310014).
文摘Numerous studies on the formation and consolidation of memory have shown that memory processes are characterized by phase-dependent and dynamic regulation.Memory retrieval,as the only representation of memory content and an active form of memory processing that induces memory reconsolidation,has attracted increasing attention in recent years.Although the molecular mechanisms specifc to memory retrievalinduced reconsolidation have been gradually revealed,an understanding of the time-dependent regulatory mechanisms of this process is still lacking.In this study,we applied a transcriptome analysis of memory retrieval at diferent time points in the recent memory stage.Diferential expression analysis and Short Time-series Expression Miner(STEM)depicting temporal gene expression patterns indicated that most diferential gene expression occurred at 48 h,and the STEM cluster showing the greatest transcriptional upregulation at 48 h demonstrated the most significant diference.We then screened the diferentially-expressed genes associated with that met the expression patterns of those cluster-identifed genes that have been reported to be involved in learning and memory processes in addition to dipeptidyl peptidase 9(DPP9).Further quantitative polymerase chain reaction verifcation and pharmacological intervention suggested that DPP9 is involved in 48-h fear memory retrieval and viral vector-mediated overexpression of DPP9 countered the 48-h retrieval-induced attenuation of fear memory.Taken together,our fndings suggest that temporal gene expression patterns are induced by recent memory retrieval and provide hitherto undocumented evidence of the role of DPP9 in the retrieval-induced reconsolidation of fear memory.
