Diabetic retinopathy (DR) is a retinal disease that causes irreversible blindness.DR occurs due to the high blood sugar level of the patient, and it is clumsy tobe detected at an early stage as no early symptoms appea...Diabetic retinopathy (DR) is a retinal disease that causes irreversible blindness.DR occurs due to the high blood sugar level of the patient, and it is clumsy tobe detected at an early stage as no early symptoms appear at the initial level. To preventblindness, early detection and regular treatment are needed. Automated detectionbased on machine intelligence may assist the ophthalmologist in examining thepatients’ condition more accurately and efficiently. The purpose of this study is toproduce an automated screening system for recognition and grading of diabetic retinopathyusing machine learning through deep transfer and representational learning.The artificial intelligence technique used is transfer learning on the deep neural network,Inception-v4. Two configuration variants of transfer learning are applied onInception-v4: Fine-tune mode and fixed feature extractor mode. Both configurationmodes have achieved decent accuracy values, but the fine-tuning method outperformsthe fixed feature extractor configuration mode. Fine-tune configuration modehas gained 96.6% accuracy in early detection of DR and 97.7% accuracy in gradingthe disease and has outperformed the state of the art methods in the relevant literature.展开更多
Objoctive To identify differential genes between normal ovarian epithelium tissue and ovarian epithelial cancer using representational difference analysis of cDNA （cDNA-RDA）. Methods cDNA-RDA was performed to ident...Objoctive To identify differential genes between normal ovarian epithelium tissue and ovarian epithelial cancer using representational difference analysis of cDNA （cDNA-RDA）. Methods cDNA-RDA was performed to identify the differentially expressed sequences between cDNAs from cancer tissue and cDNAs from normal ovarian tissue in the same patient who was in the early stage of ovarian serous cystadenocarcinoma. These differentially expressed fragments were cloned and analyzed, then sequenced and compared with known genes. Results Three differentially cxpressed cDNA fragments were isolated using cDNA from normal ovarian tissue as tester and cDNA from cancer tissue as driver amplicon by cDNA-RDA. DP Ⅲ- 1 and DP Ⅲ-2 cDNA clone showed significant homology to the cDNA of alpha actin gene; DPⅢ-3 cDNA clone showed significant homology to the cDNA oftransgelin gene. Conclusion cDNA-RDA can bc used to sensitively identify the differentially expressed genes in ovarian serous cystadenocarcinoma. Ovarian serous cystadenocarcinoma involves alteration of multiple genes.展开更多
Objective To screen coronaryartery disease (CAD) specific expressions and clone their genes. Method Blood samples were collected from CAD and non - CAD patients at the end of coronary angiography. mRNA from samples wa...Objective To screen coronaryartery disease (CAD) specific expressions and clone their genes. Method Blood samples were collected from CAD and non - CAD patients at the end of coronary angiography. mRNA from samples was isolated and converted into cDNA. After ligated with specific linkers, the cDNA was amplified with complementary primers. PCR products from CAD samples were named as tester; the ones from non - CAD samples were named as driver. With different ratio of tester to driver (1 : 100,1: 1, 000, and 1: 10, 000), they were mixed, denatured, and renatured. Single strand cD-NA was eliminated by Mung bean nuclease. Double strand cDNA presented only in tester was amplified, ligated in vector pUC19 and pUC53, and transformed into E. coll DH5a. Strains with inserted cDNA fragments were picked up based on blue and white selection. Insertions were screened by endonuclease digestion and DNA sequencing. Results were compared with DNA sequences of GeneBank. Results: After the selection with representational differential analysis, CAD specific cDNA fragments with different sizes (about 1kb, 0. 75kb, and 0. 6kb) were cloned. Among them, two fragments from unknown genes were identified. One presented a 43. 3 % similarity with part of the rattus norvegicus lipocortin gene. Another presented a 45. 4 % similarity with part of the human polynucleotide kinase 3' - phosphatase gene. Conclusion There are at least two CAD specific - ex- pressions from unknown genes that were partially similar to lipocortin and polynucleotide kinase 3'- phos-phatase genes, respectively. Expression of these genes might affect the formation and progression of plaque within coronary artery.展开更多
Understanding the neural substrates of human cognition is a central goal of neuroscience research.Modern imaging techniques,such as functional magnetic resonance imaging(fMRI),provide an opportunity to map cognitive f...Understanding the neural substrates of human cognition is a central goal of neuroscience research.Modern imaging techniques,such as functional magnetic resonance imaging(fMRI),provide an opportunity to map cognitive function in vivo.To date,modeling shared information in task-evoked neural dynamics across individuals remains challenging,largely due to pronounced inter-subject variability in brain anatomy,function,and behaviors[1],[2].An emerging topic,known as hyperalignment or functional alignment,has been proposed recently[3],to map subject-specific neural responses onto a common representational space using either linear transformations of task-evoked neural activity[4]or resting-state connectivity profiles[5].However,these approaches often assume uniform neural responses across individuals,struggling to capture group heterogeneity and model functional interactions between brain areas[6].展开更多
As a new research direction in contemporary cognitive science,predictive processing surpasses traditional computational representation and embodied cognition and has emerged as a new paradigm in cognitive science rese...As a new research direction in contemporary cognitive science,predictive processing surpasses traditional computational representation and embodied cognition and has emerged as a new paradigm in cognitive science research.The predictive processing theory advocates that the brain is a hierarchical predictive model based on Bayesian inference,and its purpose is to minimize the difference between the predicted world and the actual world,so as to minimize the prediction error.Predictive processing is therefore essentially a context-dependent model representation,an adaptive representational system designed to achieve its cognitive goals through the minimization of prediction error.展开更多
This paper presents an empirical study of the acquisition of English ambiguous verb-locative prepositional phrase constructions （VLPPs） by adult Mandarin and Spanish speakers. This study assumes that the semantic pr...This paper presents an empirical study of the acquisition of English ambiguous verb-locative prepositional phrase constructions （VLPPs） by adult Mandarin and Spanish speakers. This study assumes that the semantic properties of the target VLPPs that relate to change-of-location in sentences such as The boat floated under the bridge arise from an uninterpretable syntactic feature selected by English but unselected by Mandarin Chinese and Spanish. Results obtained from an animated cartoon selection task indicate that neither the Mandarin nor the Spanish speakers at any level of English proficiency possess native-like interpretative knowledge. Tense/ Aspect effects on the interpretation of the target constructions by Spanish speakers were also found. These results are interpreted as consistent with the Representational Deficit Hypothesis view （Hawkins, 2003, 2005） of adult second language acquisition.展开更多
Arrhythmias are a frequently occurring phenomenon in clinical practice,but how to accurately dis-tinguish subtle rhythm abnormalities remains an ongoing difficulty faced by the entire research community when conductin...Arrhythmias are a frequently occurring phenomenon in clinical practice,but how to accurately dis-tinguish subtle rhythm abnormalities remains an ongoing difficulty faced by the entire research community when conducting ECG-based studies.From a review of existing studies,two main factors appear to contribute to this problem:the uneven distribution of arrhythmia classes and the limited expressiveness of features learned by current models.To overcome these limitations,this study proposes a dual-path multimodal framework,termed DM-EHC(Dual-Path Multimodal ECG Heartbeat Classifier),for ECG-based heartbeat classification.The proposed framework links 1D ECG temporal features with 2D time–frequency features.By setting up the dual paths described above,the model can process more dimensions of feature information.The MIT-BIH arrhythmia database was selected as the baseline dataset for the experiments.Experimental results show that the proposed method outperforms single modalities and performs better for certain specific types of arrhythmias.The model achieved mean precision,recall,and F1 score of 95.14%,92.26%,and 93.65%,respectively.These results indicate that the framework is robust and has potential value in automated arrhythmia classification.展开更多
Objective To search differentially expressed sequences correlated with pathogenesis of human nasopharyngeal carcinoma (NPC), including the candidates of tumor suppressor genes Methods Representational difference a...Objective To search differentially expressed sequences correlated with pathogenesis of human nasopharyngeal carcinoma (NPC), including the candidates of tumor suppressor genes Methods Representational difference analysis (RDA) was performed to isolate differentially expressed sequences between cDNA from normal human primary cultures of nasopharyngeal epithelial cells and cDNA from NPC cell line HNE1 The source of differentially expressed products were proved by Southern blot, Northern blot and in situ hybridization The fragments were cloned with pGEM T easy kit and sequenced by the chain termination reaction Results Four differentially expressed cDNA fragments were isolated in the fourth subtractive hybridization using cDNA from normal human nasopharyngeal epithelial cells as tester amplicon and cDNA from NPC cell line HNE1 as driver amplicon by cDNA RDA These differential cDNA fragments revealed that they really came from the tester amplicon and were not expressed or down regulated in the NPC HNE1 cells Some of the genes were expressed only in human nasopharyngeal epithelial cells but deleted or down regulated in the biopsies of NPC Of these obtained clones, some were the sequences of the human known genes including house keeping genes, the others represented novel gene sequences Conclusion The differentially expressed products including the candidates of tumor suppressor genes may be associated with the initiation of the NPC展开更多
UHMWPE(Ultra-High Molecular Weight Polyethylene)plain-weave fabric,characterized by its lightweight and high-strength properties,is widely used in protective equipment such as bulletproof vests and stab-resistant vest...UHMWPE(Ultra-High Molecular Weight Polyethylene)plain-weave fabric,characterized by its lightweight and high-strength properties,is widely used in protective equipment such as bulletproof vests and stab-resistant vests,serving as a key material for enhancing protective performance.This study systematically investigates the influence mechanism of interfacial properties on the energy absorption characteristics of UHMWPE-based protective structures through multi-scale experiments and numerical simulations,and establishes a cross-scale design methodology.Innovatively,an orthotropic constitutive model incorporating dynamic friction coefficients is constructed,combined with a modified Johnson-Cook failure criterion,to achieve high-precision simulation of the entire ballistic impact process(error<3.5%).Additionally,a friction field prediction model considering strain rate effects is developed,and the friction evolution laws of UHMWPE and Para-aramid(Kevlar)fabrics under strain rates of 10^(−3) and 10^(−4) s^(−1) are obtained through MTS pull-out tests.The results show that:(1)there exists a critical yarn-yarn friction coefficient(μ=0.2);exceeding this value leads to a 19%reduction in energy absorption capacity,while viscous interfaces increase the energy dissipation peak by 16%;(2)UHMWPE shows kinetically-dominated absorption(58%)with high rate but high load,increasing damage risk.Para-aramid has friction-dominated absorption(53%)with a lower rate but stable load.Hybrid fabrics use potential-dominated absorption(49%)at a moderate rate,balancing stability and protection.(3)3–5 layers of UHMWPE fabric yield optimal cost-effectiveness,with the unit cost reduction rate of the HS+5U scheme reaching 2.74 m/(s·$),which is 91%higher than that of the hybrid scheme.(4)For HS+5U(5-ply UHMWPE),V50 is 520 m/s,meeting primary protection requirement.For hybrid solutions with U/K≥3(e.g.,HS+6U+2K),V50 reaches 580 m/s(≥540 m/s),satisfying advanced protection requirement.This research provides critical references for the design of flexible protective structures and their engineering applications.展开更多
Increased awareness of Tibetan cultural preservation,along with technological advancements,has led to significant efforts in academic research on Tibetan.However,the structural complexity of the Tibetan language and l...Increased awareness of Tibetan cultural preservation,along with technological advancements,has led to significant efforts in academic research on Tibetan.However,the structural complexity of the Tibetan language and limited labeled handwriting data impede advancements in Optical Character Recognition(OCR)and other applications.To address these challenges,this paper proposes an innovative Tibetan data augmentation technique,using Generative Adversarial Networks(GANs)to synthesise arbitrary handwriting images in variable calligraphic styles based on inputs.Moreover,our method leverages a Real-Fake Cross Inputs Strategy during training to enhance generation diversity and improve model generalisability in generating handwritten text beyond the training set and pre-defined corpus.The model was trained on three Tibetan handwriting datasets,including Ume style numerals,Uchen style consonants,and Khyug-yig style words.Experimental results demonstrate that the model successfully generates realistic and recognisable Tibetan numeral and consonant handwriting,achieving Frechet Inception Distance(FID)scores of 14.45 and 27.63,respectively.The proposed method's effectiveness in augmenting OCR models was validated as evidenced by a reduced OCR Word Error Rate(WER)on the augmented datasets.展开更多
With the recent increase in data volume and diversity,traditional text representation techniques are struggling to capture context,particularly in environments with sparse data.To address these challenges,this study p...With the recent increase in data volume and diversity,traditional text representation techniques are struggling to capture context,particularly in environments with sparse data.To address these challenges,this study proposes a new model,the Masked Joint Representation Model(MJRM).MJRM approximates the original hypothesis by leveraging multiple elements in a limited context.It dynamically adapts to changes in characteristics based on data distribution through three main components.First,masking-based representation learning,termed selective dynamic masking,integrates topic modeling and sentiment clustering to generate and train multiple instances across different data subsets,whose predictions are then aggregated with optimized weights.This design alleviates sparsity,suppresses noise,and preserves contextual structures.Second,regularization-based improvements are applied.Third,techniques for addressing sparse data are used to perform final inference.As a result,MJRM improves performance by up to 4%compared to existing AI techniques.In our experiments,we analyzed the contribution of each factor,demonstrating that masking,dynamic learning,and aggregating multiple instances complement each other to improve performance.This demonstrates that a masking-based multi-learning strategy is effective for context-aware sparse text classification,and can be useful even in challenging situations such as data shortage or data distribution variations.We expect that the approach can be extended to diverse fields such as sentiment analysis,spam filtering,and domain-specific document classification.展开更多
Distributed Denial-of-Service(DDoS)attacks pose severe threats to Industrial Control Networks(ICNs),where service disruption can cause significant economic losses and operational risks.Existing signature-based methods...Distributed Denial-of-Service(DDoS)attacks pose severe threats to Industrial Control Networks(ICNs),where service disruption can cause significant economic losses and operational risks.Existing signature-based methods are ineffective against novel attacks,and traditional machine learning models struggle to capture the complex temporal dependencies and dynamic traffic patterns inherent in ICN environments.To address these challenges,this study proposes a deep feature-driven hybrid framework that integrates Transformer,BiLSTM,and KNN to achieve accurate and robust DDoS detection.The Transformer component extracts global temporal dependencies from network traffic flows,while BiLSTM captures fine-grained sequential dynamics.The learned embeddings are then classified using an instance-based KNN layer,enhancing decision boundary precision.This cascaded architecture balances feature abstraction and locality preservation,improving both generalization and robustness.The proposed approach was evaluated on a newly collected real-time ICN traffic dataset and further validated using the public CIC-IDS2017 and Edge-IIoT datasets to demonstrate generalization.Comprehensive metrics including accuracy,precision,recall,F1-score,ROC-AUC,PR-AUC,false positive rate(FPR),and detection latency were employed.Results show that the hybrid framework achieves 98.42%accuracy with an ROC-AUC of 0.992 and FPR below 1%,outperforming baseline machine learning and deep learning models.Robustness experiments under Gaussian noise perturbations confirmed stable performance with less than 2%accuracy degradation.Moreover,detection latency remained below 2.1 ms per sample,indicating suitability for real-time ICS deployment.In summary,the proposed hybrid temporal learning and instance-based classification model offers a scalable and effective solution for DDoS detection in industrial control environments.By combining global contextual modeling,sequential learning,and instance-based refinement,the framework demonstrates strong adaptability across datasets and resilience against noise,providing practical utility for safeguarding critical infrastructure.展开更多
Object detection,a major challenge in computer vision and pattern recognition,plays a significant part in many applications,crossing artificial intelligence,face recognition,and autonomous driving.It involves focusing...Object detection,a major challenge in computer vision and pattern recognition,plays a significant part in many applications,crossing artificial intelligence,face recognition,and autonomous driving.It involves focusing on identifying the detection,localization,and categorization of targets in images.A particularly important emerging task is distinguishing real animals from toy replicas in real-time,mostly for smart camera systems in both urban and natural environments.However,that difficult task is affected by factors such as showing angle,occlusion,light intensity,variations,and texture differences.To tackle these challenges,this paper recommends Group Sparse YOLOv8(You Only Look Once version 8),an improved real-time object detection algorithm that improves YOLOv8 by integrating group sparsity regularization.This adjustment improves efficiency and accuracy while utilizing the computational costs and power consumption,including a frame selection approach.And a hybrid parallel processing method that merges pipelining with dataflow strategies to improve the performance.Established using a custom dataset of toy and real animal images along with well-known datasets,namely ImageNet,MSCOCO,and CIFAR-10/100.The combination of Group Sparsity with YOLOv8 shows high detection accuracy with lower latency.Here provides a real and resource-efficient solution for intelligent camera systems and improves real-time object detection and classification in environments,differentiating between real and toy animals.展开更多
Skin diseases affect millions worldwide.Early detection is key to preventing disfigurement,lifelong disability,or death.Dermoscopic images acquired in primary-care settings show high intra-class visual similarity and ...Skin diseases affect millions worldwide.Early detection is key to preventing disfigurement,lifelong disability,or death.Dermoscopic images acquired in primary-care settings show high intra-class visual similarity and severe class imbalance,and occasional imaging artifacts can create ambiguity for state-of-the-art convolutional neural networks(CNNs).We frame skin lesion recognition as graph-based reasoning and,to ensure fair evaluation and avoid data leakage,adopt a strict lesion-level partitioning strategy.Each image is first over-segmented using SLIC(Simple Linear Iterative Clustering)to produce perceptually homogeneous superpixels.These superpixels form the nodes of a region-adjacency graph whose edges encode spatial continuity.Node attributes are 1280-dimensional embeddings extracted with a lightweight yet expressive EfficientNet-B0 backbone,providing strong representational power at modest computational cost.The resulting graphs are processed by a five-layer Graph Attention Network(GAT)that learns to weight inter-node relationships dynamically and aggregates multi-hop context before classifying lesions into seven classes with a log-softmax output.Extensive experiments on the DermaMNIST benchmark show the proposed pipeline achieves 88.35%accuracy and 98.04%AUC,outperforming contemporary CNNs,AutoML approaches,and alternative graph neural networks.An ablation study indicates EfficientNet-B0 produces superior node descriptors compared with ResNet-18 and DenseNet,and that roughly five GAT layers strike a good balance between being too shallow and over-deep while avoiding oversmoothing.The method requires no data augmentation or external metadata,making it a drop-in upgrade for clinical computer-aided diagnosis systems.展开更多
Multimodal emotion recognition has emerged as a key research area for enabling human-centered artificial intelligence,supported by the rapid progress in vision,audio,language,and physiological modeling.Existing approa...Multimodal emotion recognition has emerged as a key research area for enabling human-centered artificial intelligence,supported by the rapid progress in vision,audio,language,and physiological modeling.Existing approaches integrate heterogeneous affective cues through diverse embedding strategies and fusion mechanisms,yet the field remains fragmented due to differences in feature alignment,temporal synchronization,modality reliability,and robustness to noise or missing inputs.This survey provides a comprehensive analysis of MER research from 2021 to 2025,consolidating advances in modality-specific representation learning,cross-modal feature construction,and early,late,and hybrid fusion paradigms.We systematically review visual,acoustic,textual,and sensor-based embeddings,highlighting howpre-trained encoders,self-supervised learning,and large languagemodels have reshaped the representational foundations ofMER.We further categorize fusion strategies by interaction depth and architectural design,examining how attention mechanisms,cross-modal transformers,adaptive gating,and multimodal large language models redefine the integration of affective signals.Finally,we summarize major benchmark datasets and evaluation metrics and discuss emerging challenges related to scalability,generalization,and interpretability.This survey aims to provide a unified perspective onmultimodal fusion for emotion recognition and to guide future research toward more coherent and generalizable multimodal affective intelligence.展开更多
The generation of high-quality 3D models from single 2D images remains challenging in terms of accuracy and completeness.Deep learning has emerged as a promising solution,offering new avenues for improvements.However,...The generation of high-quality 3D models from single 2D images remains challenging in terms of accuracy and completeness.Deep learning has emerged as a promising solution,offering new avenues for improvements.However,building models from scratch is computationally expensive and requires large datasets.This paper presents a transfer-learning-based approach for category-specific 3D reconstruction from a single 2D image.The core idea is to fine-tune a pre-trained model on specific object categories using new,unseen data,resulting in specialized versions of the model that are better adapted to reconstruct particular objects.The proposed approach utilizes a three-phase pipeline comprising image acquisition,3D reconstruction,and refinement.After ensuring the quality of the input image,a ResNet50 model is used for object recognition,directing the image to the corresponding category-specific model to generate a voxel-based representation.The voxel-based 3D model is then refined by transforming it into a detailed triangular mesh representation using the Marching Cubes algorithm and Laplacian smoothing.An experimental study,using the Pix2Vox model and the Pascal3D dataset,has been conducted to evaluate and validate the effectiveness of the proposed approach.Results demonstrate that category-specific fine-tuning of Pix2Vox significantly outperforms both the original model and the general model fine-tuned for all object categories,with substantial gains in Intersection over Union(IoU)scores.Visual assessments confirm improvements in geometric detail and surface realism.These findings indicate that combining transfer learning with category-specific fine tuning and refinement strategy of our approach leads to better-quality 3D model generation.展开更多
Accurately counting dense objects in complex and diverse backgrounds is a significant challenge in computer vision,with applications ranging from crowd counting to various other object counting tasks.To address this,w...Accurately counting dense objects in complex and diverse backgrounds is a significant challenge in computer vision,with applications ranging from crowd counting to various other object counting tasks.To address this,we propose HUANNet(High-Resolution Unified Attention Network),a convolutional neural network designed to capture both local features and rich semantic information through a high-resolution representation learning framework,while optimizing computational distribution across parallel branches.HUANNet introduces three core modules:the High-Resolution Attention Module(HRAM),which enhances feature extraction by optimizing multiresolution feature fusion;the Unified Multi-Scale Attention Module(UMAM),which integrates spatial,channel,and convolutional kernel information through an attention mechanism applied across multiple levels of the network;and the Grid-Assisted Point Matching Module(GPMM),which stabilizes and improves point-to-point matching by leveraging grid-based mechanisms.Extensive experiments show that HUANNet achieves competitive results on the ShanghaiTech Part A/B crowd counting datasets and sets new state-of-the-art performance on dense object counting datasets such as CARPK and XRAY-IECCD,demonstrating the effectiveness and versatility of HUANNet.展开更多
Dear Editor,D2This letter presents a node feature similarity preserving graph convolutional framework P G.Graph neural networks(GNNs)have garnered significant attention for their efficacy in learning graph representat...Dear Editor,D2This letter presents a node feature similarity preserving graph convolutional framework P G.Graph neural networks(GNNs)have garnered significant attention for their efficacy in learning graph representations across diverse real-world applications.展开更多
As a key carrier for the implementation of the core quality of"aesthetic creativity",the effective realization of the aesthetic value of language textbook illustrations is facing the real problems of practic...As a key carrier for the implementation of the core quality of"aesthetic creativity",the effective realization of the aesthetic value of language textbook illustrations is facing the real problems of practical disconnection and theoretical gap.Based on the theories of semiotics,embodied cognition and cultural reproduction,this paper innovatively puts forward the concept of"transformation mechanism of aesthetic education",constructs a three-phase The study analyzes the dynamic transformation process of illustrations from image symbols to students'aesthetic experience.The study explains the essential characteristics of illustration as a dual symbolic system of"likenessregulation",reveals the cognitive ladder of"perception-cognition-creation"and the path of realization of embodied cognition,and points out that the current aesthetic education of illustration has the problem of instrumentalization,and that the current aesthetic education of illustration has the problem of instrumentalization.The study points out that the current aesthetic education of illustrations is plagued by instrumental alienation,cultural hegemony and aesthetic flattening,and proposes a"three-in-one"education path from the dimensions of teachers,teaching materials and culture,including improving teachers'aesthetic education,reconstructing the principles of teaching materials and dynamic revision mechanism,and constructing classroom aesthetic education communities.The study provides new perspectives and practical guidance for exploring the aesthetic value of language textbook illustrations and bridging the gap between the theories of aesthetics and pedagogy,aiming to promote the return of aesthetic education to the nature of"reconstruction of experience",and to cultivate students'aesthetic literacy and sound personality.展开更多
In our recently published paper,[1]a typesetting error occurred during the production process.Figure 1 in the published version was incomplete.The processing of molecular dynamics(MD)simulation data into graph-structu...In our recently published paper,[1]a typesetting error occurred during the production process.Figure 1 in the published version was incomplete.The processing of molecular dynamics(MD)simulation data into graph-structured representations in the left bottom panel of thefigure was inadvertently omitted.展开更多
基金the National Research Foundation(NRF)of Korea under the auspices of the Ministry of Science and ICT,Republic of Korea(Grant No.NRF-2020R1G1A1012741)received by M.R.Bhutta.https://nrf.kird.re.kr/main.do.
文摘Diabetic retinopathy (DR) is a retinal disease that causes irreversible blindness.DR occurs due to the high blood sugar level of the patient, and it is clumsy tobe detected at an early stage as no early symptoms appear at the initial level. To preventblindness, early detection and regular treatment are needed. Automated detectionbased on machine intelligence may assist the ophthalmologist in examining thepatients’ condition more accurately and efficiently. The purpose of this study is toproduce an automated screening system for recognition and grading of diabetic retinopathyusing machine learning through deep transfer and representational learning.The artificial intelligence technique used is transfer learning on the deep neural network,Inception-v4. Two configuration variants of transfer learning are applied onInception-v4: Fine-tune mode and fixed feature extractor mode. Both configurationmodes have achieved decent accuracy values, but the fine-tuning method outperformsthe fixed feature extractor configuration mode. Fine-tune configuration modehas gained 96.6% accuracy in early detection of DR and 97.7% accuracy in gradingthe disease and has outperformed the state of the art methods in the relevant literature.
文摘Objoctive To identify differential genes between normal ovarian epithelium tissue and ovarian epithelial cancer using representational difference analysis of cDNA （cDNA-RDA）. Methods cDNA-RDA was performed to identify the differentially expressed sequences between cDNAs from cancer tissue and cDNAs from normal ovarian tissue in the same patient who was in the early stage of ovarian serous cystadenocarcinoma. These differentially expressed fragments were cloned and analyzed, then sequenced and compared with known genes. Results Three differentially cxpressed cDNA fragments were isolated using cDNA from normal ovarian tissue as tester and cDNA from cancer tissue as driver amplicon by cDNA-RDA. DP Ⅲ- 1 and DP Ⅲ-2 cDNA clone showed significant homology to the cDNA of alpha actin gene; DPⅢ-3 cDNA clone showed significant homology to the cDNA oftransgelin gene. Conclusion cDNA-RDA can bc used to sensitively identify the differentially expressed genes in ovarian serous cystadenocarcinoma. Ovarian serous cystadenocarcinoma involves alteration of multiple genes.
文摘Objective To screen coronaryartery disease (CAD) specific expressions and clone their genes. Method Blood samples were collected from CAD and non - CAD patients at the end of coronary angiography. mRNA from samples was isolated and converted into cDNA. After ligated with specific linkers, the cDNA was amplified with complementary primers. PCR products from CAD samples were named as tester; the ones from non - CAD samples were named as driver. With different ratio of tester to driver (1 : 100,1: 1, 000, and 1: 10, 000), they were mixed, denatured, and renatured. Single strand cD-NA was eliminated by Mung bean nuclease. Double strand cDNA presented only in tester was amplified, ligated in vector pUC19 and pUC53, and transformed into E. coll DH5a. Strains with inserted cDNA fragments were picked up based on blue and white selection. Insertions were screened by endonuclease digestion and DNA sequencing. Results were compared with DNA sequences of GeneBank. Results: After the selection with representational differential analysis, CAD specific cDNA fragments with different sizes (about 1kb, 0. 75kb, and 0. 6kb) were cloned. Among them, two fragments from unknown genes were identified. One presented a 43. 3 % similarity with part of the rattus norvegicus lipocortin gene. Another presented a 45. 4 % similarity with part of the human polynucleotide kinase 3' - phosphatase gene. Conclusion There are at least two CAD specific - ex- pressions from unknown genes that were partially similar to lipocortin and polynucleotide kinase 3'- phos-phatase genes, respectively. Expression of these genes might affect the formation and progression of plaque within coronary artery.
基金supported by the STI2030-Major Projects(2021ZD0200201,2022ZD0211500)the National Natural Science Foundation of China(62201519,52307259,62327805,and 82151307).
文摘Understanding the neural substrates of human cognition is a central goal of neuroscience research.Modern imaging techniques,such as functional magnetic resonance imaging(fMRI),provide an opportunity to map cognitive function in vivo.To date,modeling shared information in task-evoked neural dynamics across individuals remains challenging,largely due to pronounced inter-subject variability in brain anatomy,function,and behaviors[1],[2].An emerging topic,known as hyperalignment or functional alignment,has been proposed recently[3],to map subject-specific neural responses onto a common representational space using either linear transformations of task-evoked neural activity[4]or resting-state connectivity profiles[5].However,these approaches often assume uniform neural responses across individuals,struggling to capture group heterogeneity and model functional interactions between brain areas[6].
基金supported by the National Social Science Fund of China’s project‘Philosophical Research on the Challenge of Artificial Cognition to Natural Cognition’(grant number 21&ZD061)
文摘As a new research direction in contemporary cognitive science,predictive processing surpasses traditional computational representation and embodied cognition and has emerged as a new paradigm in cognitive science research.The predictive processing theory advocates that the brain is a hierarchical predictive model based on Bayesian inference,and its purpose is to minimize the difference between the predicted world and the actual world,so as to minimize the prediction error.Predictive processing is therefore essentially a context-dependent model representation,an adaptive representational system designed to achieve its cognitive goals through the minimization of prediction error.
文摘This paper presents an empirical study of the acquisition of English ambiguous verb-locative prepositional phrase constructions （VLPPs） by adult Mandarin and Spanish speakers. This study assumes that the semantic properties of the target VLPPs that relate to change-of-location in sentences such as The boat floated under the bridge arise from an uninterpretable syntactic feature selected by English but unselected by Mandarin Chinese and Spanish. Results obtained from an animated cartoon selection task indicate that neither the Mandarin nor the Spanish speakers at any level of English proficiency possess native-like interpretative knowledge. Tense/ Aspect effects on the interpretation of the target constructions by Spanish speakers were also found. These results are interpreted as consistent with the Representational Deficit Hypothesis view （Hawkins, 2003, 2005） of adult second language acquisition.
基金supported by the Innovative Human Resource Development for Local Intel-lectualization program through the Institute of Information&Communications Technology Planning&Evaluation(IITP)grant funded by the Korea government(MSIT)(No.IITP-2026-2020-0-01741)the research fund of Hanyang University(HY-2025-1110).
文摘Arrhythmias are a frequently occurring phenomenon in clinical practice,but how to accurately dis-tinguish subtle rhythm abnormalities remains an ongoing difficulty faced by the entire research community when conducting ECG-based studies.From a review of existing studies,two main factors appear to contribute to this problem:the uneven distribution of arrhythmia classes and the limited expressiveness of features learned by current models.To overcome these limitations,this study proposes a dual-path multimodal framework,termed DM-EHC(Dual-Path Multimodal ECG Heartbeat Classifier),for ECG-based heartbeat classification.The proposed framework links 1D ECG temporal features with 2D time–frequency features.By setting up the dual paths described above,the model can process more dimensions of feature information.The MIT-BIH arrhythmia database was selected as the baseline dataset for the experiments.Experimental results show that the proposed method outperforms single modalities and performs better for certain specific types of arrhythmias.The model achieved mean precision,recall,and F1 score of 95.14%,92.26%,and 93.65%,respectively.These results indicate that the framework is robust and has potential value in automated arrhythmia classification.
文摘Objective To search differentially expressed sequences correlated with pathogenesis of human nasopharyngeal carcinoma (NPC), including the candidates of tumor suppressor genes Methods Representational difference analysis (RDA) was performed to isolate differentially expressed sequences between cDNA from normal human primary cultures of nasopharyngeal epithelial cells and cDNA from NPC cell line HNE1 The source of differentially expressed products were proved by Southern blot, Northern blot and in situ hybridization The fragments were cloned with pGEM T easy kit and sequenced by the chain termination reaction Results Four differentially expressed cDNA fragments were isolated in the fourth subtractive hybridization using cDNA from normal human nasopharyngeal epithelial cells as tester amplicon and cDNA from NPC cell line HNE1 as driver amplicon by cDNA RDA These differential cDNA fragments revealed that they really came from the tester amplicon and were not expressed or down regulated in the NPC HNE1 cells Some of the genes were expressed only in human nasopharyngeal epithelial cells but deleted or down regulated in the biopsies of NPC Of these obtained clones, some were the sequences of the human known genes including house keeping genes, the others represented novel gene sequences Conclusion The differentially expressed products including the candidates of tumor suppressor genes may be associated with the initiation of the NPC
基金the Postdoctoral Science Foundation Funded Project of China with grant No.2021M701687Introduction and Education Plan for Young Innovative Talents in Colleges and Universities of Shandong Province.
文摘UHMWPE(Ultra-High Molecular Weight Polyethylene)plain-weave fabric,characterized by its lightweight and high-strength properties,is widely used in protective equipment such as bulletproof vests and stab-resistant vests,serving as a key material for enhancing protective performance.This study systematically investigates the influence mechanism of interfacial properties on the energy absorption characteristics of UHMWPE-based protective structures through multi-scale experiments and numerical simulations,and establishes a cross-scale design methodology.Innovatively,an orthotropic constitutive model incorporating dynamic friction coefficients is constructed,combined with a modified Johnson-Cook failure criterion,to achieve high-precision simulation of the entire ballistic impact process(error<3.5%).Additionally,a friction field prediction model considering strain rate effects is developed,and the friction evolution laws of UHMWPE and Para-aramid(Kevlar)fabrics under strain rates of 10^(−3) and 10^(−4) s^(−1) are obtained through MTS pull-out tests.The results show that:(1)there exists a critical yarn-yarn friction coefficient(μ=0.2);exceeding this value leads to a 19%reduction in energy absorption capacity,while viscous interfaces increase the energy dissipation peak by 16%;(2)UHMWPE shows kinetically-dominated absorption(58%)with high rate but high load,increasing damage risk.Para-aramid has friction-dominated absorption(53%)with a lower rate but stable load.Hybrid fabrics use potential-dominated absorption(49%)at a moderate rate,balancing stability and protection.(3)3–5 layers of UHMWPE fabric yield optimal cost-effectiveness,with the unit cost reduction rate of the HS+5U scheme reaching 2.74 m/(s·$),which is 91%higher than that of the hybrid scheme.(4)For HS+5U(5-ply UHMWPE),V50 is 520 m/s,meeting primary protection requirement.For hybrid solutions with U/K≥3(e.g.,HS+6U+2K),V50 reaches 580 m/s(≥540 m/s),satisfying advanced protection requirement.This research provides critical references for the design of flexible protective structures and their engineering applications.
文摘Increased awareness of Tibetan cultural preservation,along with technological advancements,has led to significant efforts in academic research on Tibetan.However,the structural complexity of the Tibetan language and limited labeled handwriting data impede advancements in Optical Character Recognition(OCR)and other applications.To address these challenges,this paper proposes an innovative Tibetan data augmentation technique,using Generative Adversarial Networks(GANs)to synthesise arbitrary handwriting images in variable calligraphic styles based on inputs.Moreover,our method leverages a Real-Fake Cross Inputs Strategy during training to enhance generation diversity and improve model generalisability in generating handwritten text beyond the training set and pre-defined corpus.The model was trained on three Tibetan handwriting datasets,including Ume style numerals,Uchen style consonants,and Khyug-yig style words.Experimental results demonstrate that the model successfully generates realistic and recognisable Tibetan numeral and consonant handwriting,achieving Frechet Inception Distance(FID)scores of 14.45 and 27.63,respectively.The proposed method's effectiveness in augmenting OCR models was validated as evidenced by a reduced OCR Word Error Rate(WER)on the augmented datasets.
基金supported by the SungKyunKwan University and the BK21 FOUR(Graduate School Innovation)funded by the Ministry of Education(MOE,Korea)and National Research Foundation of Korea(NRF).
文摘With the recent increase in data volume and diversity,traditional text representation techniques are struggling to capture context,particularly in environments with sparse data.To address these challenges,this study proposes a new model,the Masked Joint Representation Model(MJRM).MJRM approximates the original hypothesis by leveraging multiple elements in a limited context.It dynamically adapts to changes in characteristics based on data distribution through three main components.First,masking-based representation learning,termed selective dynamic masking,integrates topic modeling and sentiment clustering to generate and train multiple instances across different data subsets,whose predictions are then aggregated with optimized weights.This design alleviates sparsity,suppresses noise,and preserves contextual structures.Second,regularization-based improvements are applied.Third,techniques for addressing sparse data are used to perform final inference.As a result,MJRM improves performance by up to 4%compared to existing AI techniques.In our experiments,we analyzed the contribution of each factor,demonstrating that masking,dynamic learning,and aggregating multiple instances complement each other to improve performance.This demonstrates that a masking-based multi-learning strategy is effective for context-aware sparse text classification,and can be useful even in challenging situations such as data shortage or data distribution variations.We expect that the approach can be extended to diverse fields such as sentiment analysis,spam filtering,and domain-specific document classification.
基金supported by the Extral High Voltage Power Transmission Company,China Southern Power Grid Co.,Ltd.
文摘Distributed Denial-of-Service(DDoS)attacks pose severe threats to Industrial Control Networks(ICNs),where service disruption can cause significant economic losses and operational risks.Existing signature-based methods are ineffective against novel attacks,and traditional machine learning models struggle to capture the complex temporal dependencies and dynamic traffic patterns inherent in ICN environments.To address these challenges,this study proposes a deep feature-driven hybrid framework that integrates Transformer,BiLSTM,and KNN to achieve accurate and robust DDoS detection.The Transformer component extracts global temporal dependencies from network traffic flows,while BiLSTM captures fine-grained sequential dynamics.The learned embeddings are then classified using an instance-based KNN layer,enhancing decision boundary precision.This cascaded architecture balances feature abstraction and locality preservation,improving both generalization and robustness.The proposed approach was evaluated on a newly collected real-time ICN traffic dataset and further validated using the public CIC-IDS2017 and Edge-IIoT datasets to demonstrate generalization.Comprehensive metrics including accuracy,precision,recall,F1-score,ROC-AUC,PR-AUC,false positive rate(FPR),and detection latency were employed.Results show that the hybrid framework achieves 98.42%accuracy with an ROC-AUC of 0.992 and FPR below 1%,outperforming baseline machine learning and deep learning models.Robustness experiments under Gaussian noise perturbations confirmed stable performance with less than 2%accuracy degradation.Moreover,detection latency remained below 2.1 ms per sample,indicating suitability for real-time ICS deployment.In summary,the proposed hybrid temporal learning and instance-based classification model offers a scalable and effective solution for DDoS detection in industrial control environments.By combining global contextual modeling,sequential learning,and instance-based refinement,the framework demonstrates strong adaptability across datasets and resilience against noise,providing practical utility for safeguarding critical infrastructure.
文摘Object detection,a major challenge in computer vision and pattern recognition,plays a significant part in many applications,crossing artificial intelligence,face recognition,and autonomous driving.It involves focusing on identifying the detection,localization,and categorization of targets in images.A particularly important emerging task is distinguishing real animals from toy replicas in real-time,mostly for smart camera systems in both urban and natural environments.However,that difficult task is affected by factors such as showing angle,occlusion,light intensity,variations,and texture differences.To tackle these challenges,this paper recommends Group Sparse YOLOv8(You Only Look Once version 8),an improved real-time object detection algorithm that improves YOLOv8 by integrating group sparsity regularization.This adjustment improves efficiency and accuracy while utilizing the computational costs and power consumption,including a frame selection approach.And a hybrid parallel processing method that merges pipelining with dataflow strategies to improve the performance.Established using a custom dataset of toy and real animal images along with well-known datasets,namely ImageNet,MSCOCO,and CIFAR-10/100.The combination of Group Sparsity with YOLOv8 shows high detection accuracy with lower latency.Here provides a real and resource-efficient solution for intelligent camera systems and improves real-time object detection and classification in environments,differentiating between real and toy animals.
基金funded by the Deanship of Graduate Studies and Scientific Research at Jouf University under grant No.(DGSSR-2025-02-01296).
文摘Skin diseases affect millions worldwide.Early detection is key to preventing disfigurement,lifelong disability,or death.Dermoscopic images acquired in primary-care settings show high intra-class visual similarity and severe class imbalance,and occasional imaging artifacts can create ambiguity for state-of-the-art convolutional neural networks(CNNs).We frame skin lesion recognition as graph-based reasoning and,to ensure fair evaluation and avoid data leakage,adopt a strict lesion-level partitioning strategy.Each image is first over-segmented using SLIC(Simple Linear Iterative Clustering)to produce perceptually homogeneous superpixels.These superpixels form the nodes of a region-adjacency graph whose edges encode spatial continuity.Node attributes are 1280-dimensional embeddings extracted with a lightweight yet expressive EfficientNet-B0 backbone,providing strong representational power at modest computational cost.The resulting graphs are processed by a five-layer Graph Attention Network(GAT)that learns to weight inter-node relationships dynamically and aggregates multi-hop context before classifying lesions into seven classes with a log-softmax output.Extensive experiments on the DermaMNIST benchmark show the proposed pipeline achieves 88.35%accuracy and 98.04%AUC,outperforming contemporary CNNs,AutoML approaches,and alternative graph neural networks.An ablation study indicates EfficientNet-B0 produces superior node descriptors compared with ResNet-18 and DenseNet,and that roughly five GAT layers strike a good balance between being too shallow and over-deep while avoiding oversmoothing.The method requires no data augmentation or external metadata,making it a drop-in upgrade for clinical computer-aided diagnosis systems.
基金supported by the Institute of Information&Communications Technology Planning&Evaluation grant funded by the Korea government(MSIT)(No.RS-2021-II211341,AI Graduate School Support Program,Chung-Ang University)in part by the Institute of Information and Communications Technology Planning and Evaluation grant funded by the Korea government(MSIT)(Development of Integrated Development Framework that Supports Automatic Neural Network Generation and Deployment Optimized for Runtime Environment,Grant No.2021-0-00766).
文摘Multimodal emotion recognition has emerged as a key research area for enabling human-centered artificial intelligence,supported by the rapid progress in vision,audio,language,and physiological modeling.Existing approaches integrate heterogeneous affective cues through diverse embedding strategies and fusion mechanisms,yet the field remains fragmented due to differences in feature alignment,temporal synchronization,modality reliability,and robustness to noise or missing inputs.This survey provides a comprehensive analysis of MER research from 2021 to 2025,consolidating advances in modality-specific representation learning,cross-modal feature construction,and early,late,and hybrid fusion paradigms.We systematically review visual,acoustic,textual,and sensor-based embeddings,highlighting howpre-trained encoders,self-supervised learning,and large languagemodels have reshaped the representational foundations ofMER.We further categorize fusion strategies by interaction depth and architectural design,examining how attention mechanisms,cross-modal transformers,adaptive gating,and multimodal large language models redefine the integration of affective signals.Finally,we summarize major benchmark datasets and evaluation metrics and discuss emerging challenges related to scalability,generalization,and interpretability.This survey aims to provide a unified perspective onmultimodal fusion for emotion recognition and to guide future research toward more coherent and generalizable multimodal affective intelligence.
基金funded by the Research,Development,and Innovation Authority(RDIA)—Kingdom of Saudi Arabia—under supervision Energy,Industry,and Advanced Technologies Research Center,Taibah University,Madinah,Saudi Arabia with grant number(12979-iau-2023-TAU-R-3-1-EI-).
文摘The generation of high-quality 3D models from single 2D images remains challenging in terms of accuracy and completeness.Deep learning has emerged as a promising solution,offering new avenues for improvements.However,building models from scratch is computationally expensive and requires large datasets.This paper presents a transfer-learning-based approach for category-specific 3D reconstruction from a single 2D image.The core idea is to fine-tune a pre-trained model on specific object categories using new,unseen data,resulting in specialized versions of the model that are better adapted to reconstruct particular objects.The proposed approach utilizes a three-phase pipeline comprising image acquisition,3D reconstruction,and refinement.After ensuring the quality of the input image,a ResNet50 model is used for object recognition,directing the image to the corresponding category-specific model to generate a voxel-based representation.The voxel-based 3D model is then refined by transforming it into a detailed triangular mesh representation using the Marching Cubes algorithm and Laplacian smoothing.An experimental study,using the Pix2Vox model and the Pascal3D dataset,has been conducted to evaluate and validate the effectiveness of the proposed approach.Results demonstrate that category-specific fine-tuning of Pix2Vox significantly outperforms both the original model and the general model fine-tuned for all object categories,with substantial gains in Intersection over Union(IoU)scores.Visual assessments confirm improvements in geometric detail and surface realism.These findings indicate that combining transfer learning with category-specific fine tuning and refinement strategy of our approach leads to better-quality 3D model generation.
基金funded by the National Natural Science Foundation of China(62273213,62472262,62572287)Natural Science Foundation of Shandong Province(ZR2024MF144)+1 种基金Natural Science Foundation of Shandong Province for Innovation and Development Joint Funds(ZR2022LZH001)Taishan Scholarship Construction Engineering.
文摘Accurately counting dense objects in complex and diverse backgrounds is a significant challenge in computer vision,with applications ranging from crowd counting to various other object counting tasks.To address this,we propose HUANNet(High-Resolution Unified Attention Network),a convolutional neural network designed to capture both local features and rich semantic information through a high-resolution representation learning framework,while optimizing computational distribution across parallel branches.HUANNet introduces three core modules:the High-Resolution Attention Module(HRAM),which enhances feature extraction by optimizing multiresolution feature fusion;the Unified Multi-Scale Attention Module(UMAM),which integrates spatial,channel,and convolutional kernel information through an attention mechanism applied across multiple levels of the network;and the Grid-Assisted Point Matching Module(GPMM),which stabilizes and improves point-to-point matching by leveraging grid-based mechanisms.Extensive experiments show that HUANNet achieves competitive results on the ShanghaiTech Part A/B crowd counting datasets and sets new state-of-the-art performance on dense object counting datasets such as CARPK and XRAY-IECCD,demonstrating the effectiveness and versatility of HUANNet.
基金supported by the National Natural Science Foundation of China(62402399)the New Chongqing Youth Innovation Talent Project(CSTB2024NSCQ-QCXMX0035)。
文摘Dear Editor,D2This letter presents a node feature similarity preserving graph convolutional framework P G.Graph neural networks(GNNs)have garnered significant attention for their efficacy in learning graph representations across diverse real-world applications.
基金The 2026 Annual Project for the Improvement of Educational and Research Capabilities of Primary and Secondary School Teachers in Guangdong Province titled"Breaking Boundaries,Integration,and Empowerment:Innovative Practice Research on Teacher Training Courses for the Synergy of Aesthetic Education and Moral Education"(Host:Wenjiang Zhang)。
文摘As a key carrier for the implementation of the core quality of"aesthetic creativity",the effective realization of the aesthetic value of language textbook illustrations is facing the real problems of practical disconnection and theoretical gap.Based on the theories of semiotics,embodied cognition and cultural reproduction,this paper innovatively puts forward the concept of"transformation mechanism of aesthetic education",constructs a three-phase The study analyzes the dynamic transformation process of illustrations from image symbols to students'aesthetic experience.The study explains the essential characteristics of illustration as a dual symbolic system of"likenessregulation",reveals the cognitive ladder of"perception-cognition-creation"and the path of realization of embodied cognition,and points out that the current aesthetic education of illustration has the problem of instrumentalization,and that the current aesthetic education of illustration has the problem of instrumentalization.The study points out that the current aesthetic education of illustrations is plagued by instrumental alienation,cultural hegemony and aesthetic flattening,and proposes a"three-in-one"education path from the dimensions of teachers,teaching materials and culture,including improving teachers'aesthetic education,reconstructing the principles of teaching materials and dynamic revision mechanism,and constructing classroom aesthetic education communities.The study provides new perspectives and practical guidance for exploring the aesthetic value of language textbook illustrations and bridging the gap between the theories of aesthetics and pedagogy,aiming to promote the return of aesthetic education to the nature of"reconstruction of experience",and to cultivate students'aesthetic literacy and sound personality.
文摘In our recently published paper,[1]a typesetting error occurred during the production process.Figure 1 in the published version was incomplete.The processing of molecular dynamics(MD)simulation data into graph-structured representations in the left bottom panel of thefigure was inadvertently omitted.
