Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks123 from public data
- Improving Multimodal Fusion with Hierarchical Mutual Information Maximization for Multimodal Sentiment Analysis678
This work proposes a framework named MultiModal InfoMax (MMIM), which hierarchically maximizes the Mutual Information (MI) in unimodal input pairs (inter-modality) and between multimodal fusion result and unimmodal input in order to maintain task-related information through multimodAL fusion.
- 371
- A watermark strategy for quantum images based on quantum fourier transform169
A robust watermark strategy for quantum images that embeds the watermark image into the fourier coefficients of the quantum carrier image, which will not affect the carrier image’s visual effect.
- 84
- 66
- 62
- Bayesian Federated Learning: A Survey58
A critical overview of BFL is presented, including its basic concepts, its relations to Bayesian learning in the context of FL, and a taxonomy of B FL from both Bayesian and federated perspectives.
- Geometry-Contrastive Generative Adversarial Network for Facial Expression Synthesis36
A geometry-contrastive generative adversarial network GC-GAN for generating facial expression images conditioned on geometry information so that an identity-preserving face can be generated guided by the target expression.
- Single-image dehazing via depth-guided deep retinex decomposition25
A depth-guided retinex decomposition network, which consists of a boosted decoder with a fusion attention mechanism to optimize the illumination iteratively, giving rise to a refined reflectance, that could effectively realize dehazing in nighttime.
- A joint object detection and semantic segmentation model with cross-attention and inner-attention mechanisms25
This model achieves the highest performance in comparison with several recently-proposed methods and enables to build up the essential interaction between the subdividing detection branch and segmentation branch to fully make use of their correlation.
- Consolidator: Mergeable Adapter with Grouped Connections for Visual Adaptation23
Motivated by the success of group-wise convolution, grouped connections across the features extracted by fully connected layers are adopted to construct tunable parts in a consolidator to enhance the model's capacity to transfer knowledge under a constrained storage budget and keep inference efficient.
- 23
- Automated diatom searching in the digital scanning electron microscopy images of drowning cases using the deep neural networks22
A deep learning-based approach for automatically searching diatoms in scanning electron microscopic images can satisfy the necessary requirements to be integrated as part of an automatic forensic diatom test.
- 20
- Confidence-based Visual Dispersal for Few-shot Unsupervised Domain Adaptation19
The proposed C-VisDiT consists of a cross-domain visual dispersal strategy that transfers only high-confidence source knowledge for model adaptation and an intra-domain visual dispersal strategy that guides the learning of hard target samples with easy ones that significantly outperforms state-of- the-art FUDA methods.
- SANCL: Multimodal Review Helpfulness Prediction with Selective Attention and Natural Contrastive Learning19
This paper proposes SANCL: Selective Attention and Natural Contrastive Learning for MRHP, which adopts a probe-based strategy to enforce high attention weights on the regions of greater significance and constructs a contrastive learning framework based on natural matching properties in the dataset.
- 18
- 18
- An Anisotropic Gaussian Filtering Model for Image De-Hazing17
A de-hazing algorithm on the basis of the anisotropic Gaussian filtering method to overcome some essential limitations of the DCP-based (the dark channel prior, DCP) methods, such as halo artifacts and over-saturation problems is proposed.
- 17
- Convergence Analysis for the SMC‐MeMBer and SMC‐CBMeMBer Filters17
This paper proves that the SMC-MeMBer andSMC-CBMe MBer filters, respectively, converge to the true Me MBer and CB meMBer filters in the mean-square sense and the corresponding bounds for the mean ofsquare errors are given.
- Entropy-based risk network identification in adolescent self-injurious behavior using machine learning and network analysis16
Key risk factors for SIB are identified, including loneliness, ADHD symptoms, Internet addiction, anxiety, depression, affinity for solitude, autistic traits, being bullied, and the integration of EWM, network analysis, and machine learning provides a more precise risk assessment approach.
- Low-resource extraction with knowledge-aware pairwise prototype learning16
K-HPN is able to recognize inherent correlation among classes, where each class is represented as a prototype on the hypersphere and outperforms previous methods of KE, particularly with low-resource training data regimes.
- Learning to Infer Unseen Single-/ Multi-Attribute-Object Compositions With Graph Networks13
An attribute-object semantic association graph model is proposed to learn the complex relations and enable knowledge transfer between primitives and two novel metrics Hard and Soft are proposed to give a comprehensive evaluation in the multi-attribute setting.
- Transformer-Based Approach Via Contrastive Learning for Zero-Shot Detection13
The Trans-ZSD framework is proposed, which is a transformer-based multi-scale contextual detection framework that explicitly exploits inter-class correlations between seen and unseen classes and optimizes feature distribution to learn discriminative features.
- A novel image matting method using sparse manual clicks13
A new matting method based on spectral clustering, which generates a well matte using a sparse user input that outperforms most of the state-of-the-art methods with a sparseuser input and has fewer requirements to get a robust matte.
- YOLO-UniOW: Efficient Universal Open-World Object Detection12
This work introduces Universal Open-World Object Detection (Uni-OWD), a new paradigm that unifies open-vocabulary and open-world object detection tasks and designs a Wildcard Learning strategy that detects out-of-distribution objects as"unknown" while enabling dynamic vocabulary expansion without the need for incremental learning.
- Persuasive dialogue understanding: The baselines and negative results12
This paper demonstrates the limitations of a Transformer-based approach coupled with Conditional Random Field (CRF) for the task of persuasive strategy recognition and leverages inter- and intra-speaker contextual semantic features, as well as label dependencies to improve the recognition.
- View synthesis using foreground object extraction for disparity control and image inpainting12
A novel virtual view synthesis method for depth-image-based rendering to alleviate the discomfort of multiview displays that achieves effective disparity control and generates high-quality virtual view images.
- 11
- Remote Sensing Image Target Detection Method Based on Refined Feature Extraction11
This paper proposes a one-stage target detection algorithm, DCN-YOLO, based on refined feature extraction techniques, which effectively suppresses noise caused by extracting shallow features and enhances the representation of key features.
- Learning to Ask for Data-Efficient Event Argument Extraction (Student Abstract)11
This study cast EAE as a question-based cloze task and empirically analyze fixed discrete token template performance, and proposes a novel approach called “Learning to Ask,” which can learn optimized question templates for EAE without human annotations.
- De-noising ghost imaging via principal components analysis and compandor10
This optimization model may offer a promising implementation of de-noising ghost imaging by employing principal components analysis and companding technique by choosing appropriate parameters of the principal components and the companding function.
- 10
- 9
- iP3T: an interpretable multimodal time-series model for enhanced gait phase prediction in wearable exoskeletons8
The attention mechanisms within the iP3T model contribute to its interpretability, allowing for effective optimization of sensor configurations and ultimately improving mobility and quality of life for individuals with gait impairments.
- A Robust Visual Tracking Method Based on Reconstruction Patch Transformer Tracking8
This paper proposes a reconstruction patch strategy which combines the extracted features with multiple elements of the spatial dimension into a new patch which reduces the amount of network computation and the dimension of the patch sent to the transformer, thereby sharply reducing the network parameters and improving the model-tracking speed.
- Joint Generative Image Deblurring Aided by Edge Attention Prior and Dynamic Kernel Selection8
This study proposes a generative neural architecture and an edge attention algorithm developed to restore vivid multimedia patches and shows state-ofthe-art performance in terms of speed and accuracy as well as has a significant visual improvement.
- 7
- 7
- Abnormal behavior recognition based on feature fusion C3D network7
A deep learning model based on three-dimensional (3D) convolutional network multiscale feature fusion to reduce the impact of constant appearance changes, background clutter, and pedestrian occlusion is designed.
- 7
- Multi-Maneuvering Target Tracking Based on a Gaussian Process6
The presented approach delivers high-precision tracking of complex multi-maneuvering target scenarios, validating its effectiveness in addressing model uncertainty, and improves the algorithm’s adaptability in complex environments.
- Virtual View Synthesis Based on DIBR and Image Inpainting6
The method proposed in this paper can effectively remove holes and edge-ghosting and is the first method that is using morphological closing in the depth map de-noising process.
- No-reference stereoscopic image quality assessment using 3D visual saliency maps fused with three-channel convolutional neural network5
A depth-perceived 3D visual saliency map is presented and a no-reference stereoscopic image quality assessment (NR SIQA) algorithm using 3Dvisual saliency maps and convolutional neural network and CNN is proposed, which demonstrates to be competitive with the state-of-the-art NRSIQA algorithms.
- 5
- Modelling the protocol stack in NCS with deterministic and stochastic petri net5
A general modelling analysis framework for NCS is proposed based on the cross-layer concept, which is to establish an efficiency system scheduling model through abstracting the time constraint, the task interrelation, the processor and the bus sub-models from upper and lower layers (application, data link and physical layer).
- Segmenting and Understanding: Region-aware Semantic Attention for Fine-grained Image Quality Assessment with Large Language Models4
A fine-grained image quality assessment model, named RSFIQA, which integrates region-level distortion information to perceive multi-dimensional quality discrepancies and generates a global attention map by aggregating fine-grained representations from local regions is proposed.
- 4
- 4
- MiLe Loss: a New Entropy-Weighed Loss for Mitigating the Bias of Learning Difficulties in Large Language Models4
Experiments reveal that models incorporating the proposed InfoEn-tropy Loss can gain consistent performance improvement in generative language models.
- 4
- Discriminative adversarial networks for positive-unlabeled learning.4
A novel PU learning framework, called discrim inative adversarial networks, which contains two discriminative models represented by deep neural networks, that does not require the class prior estimation and its consistency can be proved under very general conditions.
- 4
- 4
- PruneHal: Reducing Hallucinations in Multi-modal Large Language Models through Adaptive KV Cache Pruning3
PruneHal is a training-free, simple yet effective method that leverages adaptive KV cache pruning to enhance the model's focus on critical visual information, thereby mitigating hallucinations, and is the first to apply token pruning for hallucination mitigation in MLLMs.
- Unsupervised single-image dehazing via self-guided inverse-retinex GAN3
An unsupervised single-image dehazing model based on the dual relationship between dehazing and Retinex is explored and a self-guided generative adversarial network (GAN) based on the dual relationship between dehazing and Retinex is proposed.
- Video super-resolution reconstruction method based on deep Back projection and motion feature fusion3
This paper proposes the idea of reconstructing key frames and non-key frames respectively, and gives a video super-resolution reconstruction method based on deep back projection and motion feature fusion, which performs better than the state-of-the-art methods, and has good robustness.
- 3
- 3
- Pharmacokinetics and tissue distribution of clevidipine and its metabolite in dogs and rats3
It was suggested that clevidipine was mainly distributed in blood and it transformed to inactive metabolite rapidly, with the highest concentrations found in the brain, heart and liver.
- 3
- 3
- A Non-Invasive Gait-Based Screening Approach for Parkinson’s Disease Using Time-Series Analysis2
The proposed approach enhances PD detection through a cost-effective, non-invasive methodology, supporting early diagnosis and disease monitoring by integrating machine learning with clinical insights, demonstrates the potential of AI-driven solutions in advancing PD screening and management.
- Enhancing image dehazing with polarization awareness and SAM-guided fusion2
This work presents a polarization-aware image dehazing method that incorporates the Segment Anything Model (SAM) for enhanced performance and leverages a polarization-inspired structural design to improve feature representation capabilities and integrates SAM to support the network’s parameter learning process.
- 2
- 2
- 2
- 2
- Saliency Object Detection Based on Regions Merging and Its Application in Image Retrieval2
A region merging strategy to solve the problem of extracting the saliency objects with complex background in the image by merging regions by means of the distance from regions to the image center and the length of the boundary.
- Double-constraint flexible tree search-based orthogonal matching pursuit for DOA estimation using dynamic sensor arrays2
The proposed approach allows exploitation of the array orientation diversity in the compressive sensing framework to address challenging array signal processing problems such as left-right ambiguity and poor estimation performance.
- Convergence Results for the Gaussian Mixture Implementation of the Extended‐Target PHD Filter and Its Extended Kalman Filtering Approximation2
This paper proves that both the GM-EP HD filter and the EK-GM-EPHD filter converge uniformly to the true EPHD filter.
- 2
- An amalgamating method of classification rules based on multiple concept lattices2
Classification rules mining from multiple concept lattices is discussed, and a method of amalgamating rules is proposed, by which all rules can be gotten, to verify this method.
- 2
- 1
- 1
- 1
- Real-time target detection method of UAV aerial image combined with X-ray digital imaging technology1
- Pose estimation of near-infrared surgical instruments based on binocular vision1
The experimental outcomes illustrated that the stability and precision of the instrument pose estimation met the demands of surgical navigation, with a root mean square error (RMSE) of 0.23 mm for dynamic tip position estimation and a maximum error of 2 degrees in the estimation of the instrument's orientation.
- 1
- 1
- 1
- An Energy-Heterogeneous Uneven Clustering Routing Algorithm for WSNs1
Simulation results show that EHUCR can prolong the network lifetime, balance network energy consumption and avoid energy holes.
- 1
- Context Awareness Based Bandwidth Management Scheme for Ad hoc Network1
Simulation results indicate that CABMS can assign more bandwidth to relatively important business under given conditions, compared with proportional fairness.
- 1
- 1
- 1
- 1
- EFFICIENTLY MINING RECENT FREQUENT PATTERNS OVER ONLINE TRANSACTIONAL DATA STREAMS1
This method uses RFP-tree to store compactly the recent frequent patterns of a stream and can reduce the average processing time of stream data element and it is superior to other analogous algorithms.
- 1
- Enhanced underwater object detection via multi-scale attention and adaptive feature fusion–
A novel model integrating multi-scale attention and adaptive feature fusion to enhance detection accuracy and robustly handles the inherent complexities of underwater environments is proposed.
- Enhancing underwater debris detection: a dimension-aware diffusion and progressive feature enhancement approach–
DRPI-DEIM is introduced, a novel lightweight debris detection model based on the DEIM framework, incorporating four architectural modules plus a distillation-based compact variant: an Information Integration Attention (IIA) mechanism for precise spatial positioning, a Progressive-Wavelet Transform Enhancement (PWTE) Module for addressing image degradation, and a RepGhost Efficient Layer Aggregation Network (RGELAN) for reduced computational cost.
- –
- –
- –
- –
- –
- –
- A Non-Invasive Gait-Based Screening Approach for Parkinson’s Disease–
This research addresses the challenges of early diagnosis of Parkinson's disease, such as the use of unreliable diagnostic techniques and limited healthcare resources, and applies camera-based data collection to analyze gait patterns that can identify a risk of Parkinson's Disease.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Classification and Comparative Analysis of Resource Management Methods in Ad Hoc Network–
Adaptive scene awareness-based resource management is more suitable for dynamic Ad hoc network and can be combined appropriately for different application scenarios in future to achieve more efficient resource management for Ad hocnetwork.
- –
- On-Demand Routing Algorithm Base on Energy Balancing in MANET–
Simulation experiments demonstrate, comparing with the LMST algorithm, the algorithm increases the average life time, decreases the average transmission power, and achieves energy balancing in the network and prolong the life-span of the network, which shows the proposed algorithm is highly practical.
- –
- An Energy-Balance Topology-Control Algorithm for Ad Hoc Networks–
The simulation experiment results show that EBDTC algorithm ensures a the network topology connectivity, extends the network lifetime so that reliable operation of the network is ensured.
- Application of Word object models in intelligent test paper–
In order to create Word documents in intelligent test paper system, Microsoft Word programing model is analyzed, and in-depth study on the object models such as Application,Document, Range and BookMark in Word is performed.
- Research on tracking of maneuvering multi-target based on bionics for IRST system–
This paper mainly studies bearing-only target tracking based on bionics for IRST system, and proposes an effective approach of state filtering algorithm with appropriate computation cost for bearing- only target in polar coordinates.
- Ant Colony Optimization and Energy Efficient Ad Hoc Routing Protocol–
Simulation results show that the algorithm can prolong the network lifetime effectively and enhance the adaptive capacity of the communication network.
- Analysis and identifying of the noises in CR images–
The conclusion of the adopted method to reduce the effect of image noise depends on the type of image which is given, which is very important for the choice of noise filter during the image restoration processing.
- Human ear recognition in three dimensions–
The experimental results on the UCR dataset of 155 subjects with 902 images under pose variations and the University of Notre Dame dataset of 302 subjects with time-lapse gallery-probe pairs are presented to demonstrate the effectiveness of the proposed algorithms and the system.
- Performances of nonlinear editing system and the choice of compressing-card–
The structure of nonlinear editing system and indexes of the main functions are introduced and an analysis and comparison of performances of the compressing-card in this system are particularly made.
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 2026-10-12. 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.