Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks235 from public data
- Highly efficient incremental estimation of Gaussian mixture models for online data stream clustering149
A probability-density-based data stream clustering approach which requires only the newly arrived data, not the entire historical data, to be saved in memory, which is highly efficient in clustering voluminous online data streams when compared to the standard EM algorithm.
- 117
- Heavy metal pollution in vegetables grown in the vicinity of a multi-metal mining area in Gejiu, China: total concentrations, speciation analysis, and health risk116
Investigation of soils and vegetables in a multi-metal mining area, Gejiu, China showed that lead and arsenic were imposing a serious threat on the local residents via vegetable consumption through vegetable consumption.
- Effect of soil washing on heavy metal removal and soil quality: A two-sided coin106
Although soil washing could effectively remove Pb and Cd in soil, it also resulted in a significant decline in soil quality, but soil neutralization could effectively alleviate the negative effects during soil washing.
- PIMNet: A Parallel, Iterative and Mimicking Network for Scene Text Recognition91
A Parallel, Iterative and Mimicking Network (PIMNet) to balance accuracy and efficiency is proposed, which adopts a parallel attention mechanism to predict the text faster and an iterative generation mechanism to make the predictions more accurate.
- 88
- Heavy metal pollution and potential health risks of commercially available Chinese herbal medicines85
The results showed that the Cd, Hg and Cu concentrations in the 60 CHMs exceeded the limiting values of the "Green Trade Standards of Importing & Exporting Medicinal Plants & Preparations" (WM2-2001), with exceedance ratios of 38.8%, 8.3% and 1.7%, respectively, indicating that Cd had relatively low bioactivity and toxicity.
- 53
- NTIRE 2022 Challenge on High Dynamic Range Imaging: Methods and Results45
This paper reviews the challenge on constrained high dynamic range (HDR) imaging that was part of the New Trends in Image Restoration and Enhancement (NTIRE) workshop, held in conjunction with CVPR 2022, and the competition set-up, datasets, proposed methods and their results.
- Mask is All You Need41
This work argues that the performance degradation results from the learning confusion issue in the mask head, and proposes instance-aware mask learning in which the Mask R-CNN head learns to predict the shape of the whole instance rather than classify each pixel to text or non-text.
- 40
- Multimodal sentiment analysis based on cross-instance graph neural networks39
A cross-instance graph neural network is proposed that leverages the global characteristics of the dataset to detect sentiments in text-image pairs and a multimodal in-deep fusion with the multihead attention mechanism is implemented to better predict the sentiment of image-text pairs.
- 39
- The order effect in human abductive reasoning: an empirical and computational study36
It is concluded that when little is known about the uncertainty and the dynamics of the environment, the order effect results from one's coherently and dynamically adaptive expectations of the statistical properties of the environments.
- Indole-3-acetic acid promotes cadmium (Cd) accumulation in a Cd hyperaccumulator and a non-hyperaccumulator by different physiological responses33
IAA promotes Cd accumulation in plant shoots by enhancing the accumulation of carotenoids and proline in S. nigrum and maintaining a high leaf SOD activity in S .
- 31
- Graph Convolutional Network Based on Multi-Head Pooling for Short Text Classification31
A Multi-head-Pooling-based Graph Convolutional Network (MP-GCN) for semi-supervised short text classification is proposed, and its three architectures, which focus on the node representation learning of 1-order, 1&2-order of isomorphic graphs, and 1- order of heterogeneous graphs, are introduced.
- 30
- 30
- 30
- Particle swarm optimization service composition algorithm based on prior knowledge29
According to the analysis of the experimental results, the improved particle swarm service composition method proposed in this paper shows good performance in three aspects: optimality, time complexity and convergence.
- 28
- Multi-Party Empathetic Dialogue Generation: A New Task for Dialog Systems28
A Static-Dynamic model for Multi-Party Empathetic Dialogue Generation,SDMPED, is introduced as a baseline by exploring the static sensibility and dynamic emotion for the multi-party empathetic dialogue learning, the aspects that help SDMPED achieve the state-of-the-art performance.
- A Span-based Multimodal Variational Autoencoder for Semi-supervised Multimodal Named Entity Recognition27
Experimental results demonstrate that the proposed novel span-based multimodal variational autoencoder (SMVAE) model not only outperforms baselines under supervised setting, but also improves MNER performance with less labeled data than existing semi-supervised methods.
- 26
- Effects of kinetin on plant growth and chloroplast ultrastructure of two Pteris species under arsenate stress25
It is suggested that kinetin could maintain and reorganize the ultrastructure integrality of As-stressed chloroplasts to some extent for the two plants, especially at high concentration.
- Optimization of Topic Recognition Model for News Texts Based on LDA21
This method transforms the traditional topic cluster number selection problem into clustering problem and thus can be used to optimize the topic recognition model for news texts, and does not need iterative optimization and can simplify model development.
- 20
- 19
- 19
- Effects of kinetin on arsenic speciation and antioxidative enzymes in fronds of the arsenic hyperaccumulator Pteris cretica var. nervosa and non-hyperaccumulator Pteris ensiformis19
Investigation of the effects of kinetin (KT, a type of CTKs) on arsenic (As) speciation and antioxidative enzymes in plant fronds found it promoted As transport and increased frond POD activity in P. nervosa, which contributed to its As tolerance under stress from different As species.
- Fast convergence of regularised Region-based Mixture of Gaussians for dynamic background modelling19
An expression is derived to prove the O ( 1 / k 2 ) convergence rate of the online gradient method, with momentum type updates, when the individual gradients are constrained by a growth condition.
- 18
- MIPI 2023 Challenge on Nighttime Flare Removal: Methods and Results18
The developed solutions in this challenge achieved state-of-the-art performance on Nighttime Flare Removal and a detailed description of all models developed is provided in this paper.
- Variations of arsenic forms and the role of arsenate reductase in three hydrophytes exposed to different arsenic species18
As accumulation mainly occurred via biological uptake rather than physicochemical adsorption, and AR played an important role in As detoxification in aquatic plants.
- 18
- Multimodal Brain-Computer Interfaces: AI-powered Decoding Methodologies17
This review highlights the core decoding algorithms that enable multimodal BCIs, including a dissection of the elements, a unified view of diversified approaches, and a comprehensive analysis of the present state of the field, to provide a comprehensive understanding for AI-powered multimodal BCIs.
- 17
- 17
- 17
- 16
- Topic-Aware Fake News Detection Based on Heterogeneous Graph16
A topic-aware fake news detection (FND) method based on heterogeneous graphs, the model investigates the effect of news topics onfake news detection and enhances the discriminative ability ofFake news detection.
- 16
- Chinese Implicit Sentiment Analysis Based on Hierarchical Knowledge Enhancement and Multi-Pooling16
A novel model based on “Hierarchical Knowledge Enhancement and Multi-pooling” (HKEM), which fully integrates the knowledge information of different levels in the text through hierarchical knowledge enhancement to alleviate the “weak feature” problem.
- Perception of Space by Multiple Intrinsic Frames of Reference16
Two experiments reported here suggest that distortion and segmentation arise due to a hierarchical organization of the spatial representation that is more strongly encoded and easier to recall than those associated with non-salient landmarks.
- TextBlock15
A pioneering network termed TextBlock is developed, and a heuristic text block generation method as well as a multi-instance block-level recognition module are proposed, which are expected to have a potential impact on scene text spotting research in the future.
- Incorporating Circumstances into Narrative Event Prediction15
This work proposes the CircEvent, which adopts the multi-head attention to retrieve circumstances at the local and global levels, and introduces a regularization of attention weights to leverage the alignment between events and local circumstances.
- Unsupervised Keyword Extraction Methods Based on a Word Graph Network15
An unsupervised keyword extraction method based on word graph networks for both Chinese and English is proposed that uses word embedding to applying a “word attraction score” to semantic relevance between words in a document.
- A significant positive correlation between endogenous trans-zeatin content and total arsenic in arsenic hyperaccumulator Pteris cretica var. nervosa14
A higher endogenous Z content contributed to As accumulation of P. nervosa under As stress, and a significant positive correlation was found between the contents of endogenous Z and total As in P. ensiformis.
- 13
- 12
- Object detection based on semi-supervised domain adaptation for imbalanced domain resources12
A semi-supervised detector that adapts the domain shifts on both appearance and semantic levels is proposed and the results verify the higher accuracy and effectiveness of the proposed detector in comparison with state-of-the-art detectors.
- 11
- Endogenous trans-zeatin content in plants with different metal-accumulating ability: a field survey11
Significant correlations were observed between leaf trans-zeatin content in Pteris vittata and As, Pb, and Cd concentrations in the soil; however, either no correlation or a negative one was observed in the other plant species.
- Prioritized Spatial Updating in the Intrinsic Frame of Reference11
It was found that given a relatively stable intrinsic reference direction, responses to a small number of salient objects were faster than responses to non-salient objects and a type of orientation dependence similar to that found in egocentric spatial updating was revealed.
- 10
- 10
- Efficient Progressive High Dynamic Range Image Restoration via Attention and Alignment Network10
A lightweight neural network called Efficient Attention-and-alignment-guided Progressive Network (EAPNet) for the challenge NTIRE 2022 HDR Track 1 and Track 2 is proposed and a multi-scale lightweight encoding module to extract features is introduced.
- A review of the aromatic genus Adenosma: Geographical distribution, traditional uses, phytochemistry and biological activities10
A critical appraisal of the current state of knowledge regarding the geographical distribution, traditional uses, phytochemicals and biological properties of Adenosma spp.
- Person re-identification by integrating metric learning and support vector machine10
A joint learning method by integrating the discriminative metric learning, the support vector machine (SVM) and the identity discriminator into one model, so as to realize joint construction of metric learning and identity discriminators is proposed.
- Cognitive-Affective Interactions in Human decision-Making: A Neurocomputational Approach - eScholarship10
This paper attempts to explore human cognitive-affective interactions in strategic thinking through a neurocomputational modeling approach and to empirically justify the model, carried out a neuroimaging experiment using the ultimatum game.
- Cognitive-Affective Interactions in Human decision-Making: A Neurocomputational Approach10
This paper attempts to explore human cognitive-affective interactions in strategic thinking through a neurocomputational modeling approach and to empirically justify the model, carried out a neuroimaging experiment using the ultimatum game.
- Improving autoencoder by mutual information maximization and shuffle attention for novelty detection9
A novelty detection model based on shuffle attention mechanism and mutual information maximization (MIM) to modify the effect of traditional AE on the reconstruction of inlier and outlier samples is proposed.
- Clustering of Short Texts Based on Dynamic Adjustment for Contrastive Learning9
A clustering method based on Dynamic Adjustment for Contrastive Learning (DACL), which smoothly transitions loss weight of model from Contrastive learning to clustering during training and filters negative samples in contrastive learning by the pseudo-labels generated by clustering.
- Modeling of complex internal logic for knowledge base completion9
This paper proposes the modeling of complex internal logic for knowledge base completion by integrating the semantic information into the knowledge representation model and strengthens the credibility scores of the positive and negative triples with the semantic gap, which not only makes the model converge faster, but also can obtain theknowledge representation of the fusion semantic information.
- Semantic Feature Graph Consistency with Contrastive Cluster Assignments for Multilingual Document Clustering8
The proposed SFGC3A method implements consistency objective and reconstruction objective in different feature spaces, thus effectively avoiding conflicts between consistency learning and inconsistency reconstruction and design the semantic feature graph consistency and semantic label consistency modules to further explore consistent semantic information among multilingual documents.
- A Novel Approach of Calculating Information Entropy in Information Extraction8
A novel calculation method of mixing entropy is presented, which can more accurately reflect the topic information of web page and shows great superiority over other traditional methods in both the accuracy of information entropy calculation and discrimination between topic- related information blocks and topic-unrelated information blocks.
- EigenBody: Analysis of body shape for gender from noisy images8
The paper confirms that intuitive thoughts about what properties of the human body are important for gender classification, can be effectively represented in a small number of key components and that eigenpeople reproduced using just these key components can be visually classified by gender to a large extent.
- 7
- 7
- 7
- Enhancing Semi-Supervised Learning with Cross-Modal Knowledge7
This paper proposes a novel method to further enhance SSL by introducing semantic modal knowledge, which contains the word embeddings of class labels and the semantic hierarchy structure among classes, and thus improves the generalization ability of the model.
- Effects of Oxalic Acid on Arsenic Uptake and the Physiological Responses of Hydrilla verticillata Exposed to Different Forms of Arsenic7
The results may be useful for the phytoremediation of waste water containing As and OA, but it depends on OA concentration and the form of As.
- Effect of arsenate on endogenous levels of cytokinins with different existing forms in two Pteris species7
The CTK forms in the As hyperaccumulator Pteris cretica var.
- 7
- A Clustering Algorithm Based on the Detection of Density Peaks and the Interaction Degree Between Clusters6
A two-phase clustering algorithm based on detecting the peaks of dimensional density and the degree of interaction between clusters (CPDD-ID) and a new similarity measure based on the interaction degree between clusters is proposed in the merging stage.
- 6
- 6
- Task-Specific Data Augmentation and Inference Processing for VIPriors Instance Segmentation Challenge6
Experimental results demonstrate that proposed method can achieve a competitive result on the test set of 2022 VIPriors Instance Segmentation Challenge, with 0.531 AP@0.50:0.95.
- Graph isomorphism identification based on link-assortment adjacency matrix6
New concepts such as binary link path and link-assortment adjacency matrix are introduced and a new method is proposed to improve the operability of isomorphism identification and relieve the computational pressure.
- A Neural Joint Model with BERT for Burmese Syllable Segmentation, Word Segmentation, and POS Tagging6
This study extends the neural joint model for Vietnamese word segmentation, POS tagging, and dependency parsing with the pre-training method of the Burmese character, syllable, and word embedding with BiLSTM-CRF-based neural layers and shows that the proposed joint model can result in an excellent performance.
- A Dialogue-based Information Extraction System for Medical Insurance Assessment6
A dialogue-based information extraction system that integrates advanced NLP technologies for medical insurance assessment that reduces the average time cost of the procedure from 55 minutes to 35 minutes and the total human resources cost is saved 30% compared with the previous offline procedure.
- A comparative proteomics method for multiple samples based on a 18 O-reference strategy and a quantitation and identification-decoupled strategy6
A method combining 18O-reference strategy and a quantitation and identification-decoupled strategy that made protein identification possible for all samples using a single pooled sample, and therefore gave a good reproducibility in protein identification across multiple samples, and allowed for optimizing peptide identification separately so as to identify more proteins.
- Code-based authentication with designated verifier6
A new strong DVS scheme based on CFS signature and Stern's identification scheme, whose security depends on syndrome decoding problem and bounded decoding problem, and satisfies the strong designated verifier, unforgeability and non-delegatability properties which are the security requirements for a strong D VS.
- 6
- Detecting Low Complexity Clusters by Skewness and Kurtosis in Data Stream Clustering.6
This work utilizes tests for normality based on skewness and kurtosis to discover cluster pairs that can be merged to produce a less complex normal cluster even if they have different means or covariance structures.
- 6
- 5
- System log anomaly detection based on contrastive learning and retrieval augmented5
A log anomaly detection framework named LogSentry based on contrastive learning and retrieval-augmented based on contrastive learning and retrieval-augmented achieves high performance over baseline methods.
- 5
- 5
- 5
- Ancient Chinese Sentence Segmentation Based on Bidirectional LSTM+CRF Model5
A novel method for the segmentation of Archaic Chinese sentences based on a bidirectional long short-term memory (LSTM) + conditional random field (CRF) model that introduced the stochastic gradient descent (SGD) to prevent excessive fitting and the viterbi algorithm was used to calculate the optimal sequence of the sentences.
- Word Segmentation for Burmese Based on Dual-Layer CRFs5
A word segmentation method with fusion conditions of double syllable features is proposed, which combinesword segmentation and segmentation of syllables into one process, thus reducing the impact of errors on the syllable segmentations of Burmese.
- 5
- Cross-Target Stance Detection with Multi-Level Information Fusion4
This approach, which integrates word-level and sentence-level information for cross-target stance detection, not only integrates word-level and sentence-level information for cross-target stance detection but also captures subtle semantic differences of key terms within a text.
- Co-learning Graph Convolution Network for Mobile User Profiling4
This work proposes a co-learning graph convolutional network (C-GCN) based on selective-scale attention (SS-Attention) to realize the extraction of spatial features of graphs in order to deeply understand the mobile users.
- 4
- Representation Learning with LDA Models for Entity Disambiguation in Specific Domains4
A domain entity disambiguation method that fuses the doc2vec and LDA topic models that outperforms the existing model, which proves its feasibility and effectiveness.
- Design of a System to Generate a Four Quadrant Signal at High-Frequency4
The project finished by doing research for some available signal generators, like Phase-Locked Loop (PLL) and Direct Digital Synthesis (DDS), and the final system with high frequency output signal has been designed after the research.
- 4
- 4
- Advances in the Study of Effects of Arsenic on Plant Photosynthesis4
A review was presented of the advances in and prospects of the study of effects of arsenic on plant photosynthesis from the angle of external symptoms of arsenic phytotoxicity,chloroplast ultrastructure, photoynthetic pigment, photosynthetic rate and photosynthesis process.
- A hybrid system of abductive tactical decision making4
A hybrid system to simulate this process of intelligent decision making, which combines a symbolic Soar component for deliberate explanation generation and a connectionist UEcho component for implicit explanation evaluation.
- A multilevel approach to modeling human cognition4
A hybrid system can score high in the Newell Test but may not offer a veridical and coherent theory of cognition.
- Supervised Consensus Anchor Graph Hashing for Cross Modal Retrieval3
In SCAGH, the anchor points are automatically selected and consensus anchor graph learning is integrated in an unified framework, and the anchor graph matrix is utilized to approximate the pairwise similarity matrix so that the high complexity can be avoided and enhance the quality of hash codes.
- A Masked Self-Supervised Pretraining Method for Face Parsing3
A self-supervised learning method attempting to make full use of the unlabeled facial images for face parsing by randomly mask some patches in the central area of facial images, and the model is required to reconstruct the masked patches.
- Dual Data Augmentation Method for Data-Deficient and Occluded Instance Segmentation3
Experimental results demonstrate that proposed DDA method can achieve a competitive result on the DeepSportRadar challenge, with 0.757AP@0.50:0.95 on the challenge set.
- Calculation method of formation dip angle based on deep learning3
This paper proposes a method for calculating the formation dip angle using deep learning and shows that this method more realistically reflects the undulating characteristics of a structure with both high accuracy and anti-interference ability.
- The Second Place Solution for The 4th Large-scale Video Object Segmentation Challenge--Track 3: Referring Video Object Segmentation3
This work proposes several tricks to boost the referring video object segmentation task, including cyclical learning rates, semi-supervised approach, and test-time augmentation inference, based on the strong baseline framework--ReferFormer.
- 3
- Generating Functions in Neural Learning of Sequential Structures.3
This work presents a unique method to derive the generating functions for the waiting time of sequential patterns, then compares these functions with the neural mechanisms for learning sequential structures to show that the way the neocortex integrates information over time bears a striking resemblance to the way these normative functions operate.
- On Quantum Models of the Human Mind3
Why the two levels of commitment, treating the human brain as a quantum computer and merely adopting abstract quantum probability principles to model human cognition, should be integrated are discussed.
- 3
- A Speech Based Approach to Surveillance Video Retrieval3
The anatomy of a pilot surveillance system with a speech-based interface for content-based retrieval of video data and an ontology-based information sharing architecture that lets components of the system communicate among each other through TCP/IP communication channels is described.
- 3
- Skill Acquisition: Models3
Skill acquisition is the process whereby a learner progresses from slow, deliberate, memory-intensive, error-prone performance on a task, to rapid, automatic, near-error-free performance.
- 3
- 2
- KANAD: Topologically Adaptive Graph Feature Learning for Multimodal Anomaly Detection in Microservice Systems2
A novel semi-supervised anomaly detection framework, named KANAD, which integrates fine-grained modeling of multi-modal structural and multi-modal fusion module, and introduces multi-modal fusion module to enhance the model’s generalization ability in identifying anomalous data.
- Clustering Algorithm by Boundary Detection Base on Entropy of KNN2
This work proposes a clustering algorithm by boundary detection based on entropy of KNN (CBDEK) that is proven by a comparative evaluation of nine algorithms on 24 datasets and incorporates noise mitigation techniques using average distance and box plot analysis.
- Document-Level Relation Extraction with Uncertainty Pseudo-Label Selection and Hard-Sample Focal Loss2
This work proposes the DocRE method based on uncertainty pseudo-label selection, which first trains a teacher model to annotate pseudo-labels for a dataset with incomplete annotations, trains a student model on the dataset with annotated pseudo-labels, and uses the trained student model to predict relations on the test set.
- Robot Path Planning Based on Improved Reinforcement Learning2
This paper proposes a path planning method for mobile robots based on improved reinforcement learning, which establishes a grid environment model, defines the return value through the number of steps of the robot, and proposes a changing action selection strategy for the balance between the robot’s exploration and utilization of the environment in reinforcement learning.
- Strong Instance Segmentation Pipeline for MMSports Challenge2
A strong instance segmentation pipeline is designed that can achieve a competitive result on the DeepSportRadar challenge, with 0.768AP@0.50:0.95 on the challenge set.
- IIAS: An Intelligent Insurance Assessment System through Online Real-time Conversation Analysis2
An Intelligent Insurance Assessment System (IIAS) is proposed that helps effectively collect claimant information through online real-time conversation analysis and the average time cost of the insurance assessment procedure is reduced from 55 minutes to 35 minutes.
- 2
- A Parameter Transfer Method for HMM-DNN Heterogeneous Model with the Scarce Mongolian Data Set2
The concept of the homogeneous model heterogeneous model is defined and a parameter learning method for HMM-DNN heterogeneity model in the scarce Mongolian data set is proposed, which achieves much better performance on Mongolian recognition accuracy.
- Learning Temporal Structures of Random Patterns by Generating Functions2
Using segmentation of pattern events and diagrams of Markov chains to illustrate the recursive structures represented by generating functions, a method of generating functions is presented to compute the distributions of the first-arrival and inter-Arrival times of random patterns in independent Bernoulli trials and first-order Markov trials.
- 2
- 2
- Bottom-Up Recognition Learning: A Compilation-Based Model of Limited-Lookahead Learning2
Bottom-Up Recognition Learning is described, a model of limited-lookahead learning based on final first learning and knowledge compilation that shows that BURL must be used in conjunction with other learning methods to fully explain skill acquisition on limitedlookahead tasks.
- 2
- Improve Mongolian-Chinese translation by Introducing SMT Information into NMT2
The experiment results show that applying the improved accuracy of statistic alignment of phrase-based statistic machine translation (PBSMT) to Mongolian-Chinese translation demonstrates promising improvements.
- 2
- Thai Language Sentence Similarity Computation Based on Syntactic Structure and Semantic Vector2
This paper proposes a novel method to compute the similarity of Thai language sentence based on syntactic structure and semantic vector, which firstly uses the Part-of-Speech (POS) dependency, and then through the word vector to calculate two sentences semantic similarity.
- 2
- Research on the key technologies of corpus preprocessing in Mongolian-Chinese SMT2
The key technology of case processing in the Mongolian is studied to alleviate data sparse problem of corpus effectively, which improves the performance of SMT.
- A Syntax Parsing Method Based on Adaptive Genetic Annealing Optimization HMM2
An improved syntax parsing method based on adaptive genetic annealing and HMM was presented, and the experiment results prove that this method is very effective for syntactic parsing.
- 2
- Cyber Trust and Suspicion: A Human-Centric Approach2
In this position paper, it is argued for the need of studying cyber trust and suspicion from a human-centric approach and, based on a principled abduction-based framework, the results will answer a full range of fundamental questions regarding cyberTrust and suspicion.
- The Techniques of Eye Movement Operating System Based on Human Eye Movements Characteristics2
Because of the significant performance of the eye movement operating system in the human-computer interaction and military area, some typical applications were introduced firstly and several key technical parameters in the eye tracking technology and active control input section were analyzed respectively.
- 2
- 2
- 2
- 2
- Representing is more than emulating2
To explain cognitive performance in a specific task, a systematic analysis of the underlying representational structures and their interactions is needed.
- 2
- Zero-shot stance detection incorporating sentiment knowledge in a joint contrastive learning framework1
A novel zero-shot stance detection framework that integrates sentiment knowledge into a joint contrastive learning architecture and explicitly models their interaction through a dual dynamic target-aware graph and a dual-channel contrastive learning mechanism, enabling more robust transfer of stance semantics to unseen targets.
- 1
- 1
- 1
- 1
- 1
- 1
- COSSUM: Towards Conversation-Oriented Structured Summarization for Automatic Medical Insurance Assessment1
The task of conversation-oriented structured summarization which aims to automatically produce the desired structured summary from a conversation automatically is proposed, and the results show the effectiveness of the proposed approach.
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- Key Research of Pre-processing on Mongolian-Chinese Neural Machine Translation1
This paper successfully build an attention-based neural machine translation to do the CWMT2009 Mongolian to Chinese translation task, and uses four different approaches toDo the pre-processing for both Mongolian and Chinese, including segmenting Chinese into character and separating the Mongolian stem from the suffixes.
- 1
- Transformer Defect Data Analysis Based on Data Mining Technology1
By analyzing a large number of main transformer operation defect data mining, this passage finds the fault conditions of various types of transformers obtained under various regions and in different climates, and provides reference and guidance for the selection and maintenance of transformer in different regions and climates.
- 1
- Effects of Arsenic on Physiological and Biochemical Properties of Pteris cretica var. nervosa Under Two Culture Conditions1
It was concluded that free proline and soluble sugar might play an important role in As tolerance of Pteris cretica var.nervosa.
- Acquisition of Domain Ontology from Relational Schemes1
This paper analyses how to identify the information about the structure of relational schemes in legacy systems, presents 12 transformation rules, according to which terms and relations between them can be obtained from the relational scheme in reverse way, and uses the extension entity relationship (EER) diagram to further get semantic information from relational schemes for refining ontology model.
- 1
- Content-based TV sports video retrieval using multimodal analysis1
The paper first defines the basic structure of sports video database system, and then introduces a new approach that integrates visual stream analysis, speech recognition, speech signal processing and text extraction to realize video retrieval.
- Declarative and Procedural Learning in Alphabetic Retrieval1
This paper presents three experiments that study declara- tive and procedural learning in alphabetic retrieval based on the view that speed-up during skill acquisition can result from acquiring either new procedural knowledge or new declarative knowledge, followed by speed up of both types of knowledge.
- –
- scHCL-SDA: A Homophily-Aware Contrastive Learning with Spectral Distribution Alignment for scRNA-seq Data Clustering–
A homophily-aware contrastive learning with spectral distribution alignment (scHCL-SDA) for clustering scRNA-seq data and introduces an adaptive homophily-aware consistency contrastive learning mechanism to dynamically construct reliable neighborhoods, thereby effectively blueucing structural noise in low-homophily graphs.
- –
- –
- HyRES: Measuring Node Influence via Hyperspherical Representation Equilibrium Shift–
This work proposes Hyperspherical Representation Equilibrium Shift (HyRES), a theoretically grounded, unsupervised framework that redefines node influence as a geometric displacement within the latent space, and leverages Linear Response Theory to derive a closed-form approximation of the global representation shift caused by a node's removal.
- Dual-expert collaborative model for cross-lingual cross-target stance detection–
The Dual-Expert Collaborative Model (DECM) is proposed, a unified framework that synergizes text-target interaction modeling with cross-lingual adaptation and target distribution alignment through three designed components that significantly outperforms baseline methods across multiple cross-lingual and cross-target stance detection benchmarks.
- –
- –
- An External Knowledge-Guided Generative Model for Few-Shot Aspect Category Sentiment Analysis–
A generative ACSA model that incorporates external knowledge is proposed, which retrieves fine-grained aspect-related terms from external knowledge bases and uses them to construct contrastive sentence pairs, generating enhanced aspect-related representations, thereby bridging the gap between the original text and predefined aspect categories.
- –
- –
- –
- –
- –
- Micro-cluster Structure Clustering Based on Weight-Constrained Minimum Spanning Tree–
This paper proposes a Micro-cluster Structure Clustering algorithm based on the Weight-constrained Minimum Spanning Tree, called MSC-WMST, which exhibits excellent performance in three evaluation metrics.
- –
- –
- –
- –
- –
- –
- A combined application of Deep Learning Recon and MRI mute technique–
A DL-Recon post-processing model is developed to enhance SNR and clarity of images which are acquired from optimized knee scanning protocol with mute technique and enhances the success of clinical scans by decreasing scan noise to improve patient comfort.
- –
- Robust zero-shot discrete hashing with noisy labels for cross-modal retrieval–
This work proposes a robust zero-shot discrete hashing with noisy labels (RZSDH) method, which fully considers the impact of noisy labels in real scenes and incorporates the sparse and low-rank constraints on the noise matrix and the recovered label matrix to effectively reduce the negative impact of noisy labels.
- Hypercube Pooling for Visual Semantic Embedding–
The proposed Hypercube Pooling (HCP) aggregator, which concatenates the Max and Min Pooling features as the global features, has excellent performance with high efficiency, confirming the correctness of the spatial analysis and the effectiveness of the HCP aggregator.
- –
- –
- –
- –
- –
- Insulator Pollution Component Identification Method Based on Multi-scale 3D Convolution Network Mode–
- Texture and semantic integrated small objects detection in foggy scenes–
A texture and semantic integrated small objects detection in foggy scenes that focuses on extracting discriminative features unaffected by the environment, and obtaining texture information and high-level semantic information of small objects.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Fusion News Elements of News Text Similarity Calculation–
The proposed method fully integrates the influence of the five news feature words into the evaluation of text similarity, which avoids the problem happened in the text interference and computational efficiency to a large extent.
- –
- –
- Research on Detection Algorithm for Rail Fastener Based on Computer Vision–
A detection algorithm for rail fastener based on computer vision is proposed, which combines projection method and scanning of pixels in specific region to position the position of fastener, and adopts gray-scale features and HOG features to describe feature vector offastener.
- –
- –
- –
- –
- –
- Reply to Aksentijevic: It is a matter of what is countable and how neurons learn–
It is recognized that the authors did not fudge their random sequences to create structures where none actually exist, and differences are mathematical facts that are readily observed and replicable by anyone.
- A Classification Algorithm for Chinese Verb Phrases Using Support Vector Machine–
Using 3500 sentences to train and test, experiment results show that the SVM model dramatically reduces the training time and steps and fully illustrates that the performance of the proposed algorithm is superior classification algorithm.
- On provably secure code-based multiple grade proxy signature scheme–
A code-based multiple grade proxy signatures based on Stern's identification scheme is introduced, which is secure on the syndrome decoding problem and satisfies verifiability, unforgeability, undeniability, non-transferability and distinguishability properties, which are the security requirements for a multiple grade proxy signature.
- CAPTCHA Recognition Based on Markov field–
To decrease uncertain recognition of CAPTCHA, the authors need to use a random model of using the whole info of the picture, and with using this theory, they can find out thatCAPTCHA have a good Image segementation.
- An Improved Method for Deep Web Sources Classification Based on the Theme and Form Attributes–
A novel classification method and an improved similarity measure of query interfaces were proposed to realize the automatic classification of large masses of deep web sources, which make full use of theme information and form attributes.
- –
- Head-shoulder detection under complex background based on Joint HOG features–
The Haar-like classifier works very fast, and can reject most negatives in the first stage of detection, while the proposed feature Joint HOG, which can well describe the symmetry of an-like shape, selects the true positives again.
- –
- –
- Design and Implementation of IIR Digital Filter Based on Matlab–
This paper proposes one rapid and effective design scheme using Matlab toolbox FDATOOL to realize IIR digital filter design proposal, which can adjust the parameter of filter and thus optimize the filter design.
- Research of Algorithm of Chinese Verb Phrases Categorization Based on Support Vector Machine–
Experiments show that the SVM model dramatically reduces the training time and steps and is a better classification algorithm.
- Design of searching engine based on SVM and statistics–
The result indicates advantage and development of the searching engine based on support vector machine and statistics.
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-15. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.