Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks79 from public data
- LanguageBind: Extending Video-Language Pretraining to N-modality by Language-based Semantic Alignment490
This work proposes LanguageBind, taking the language as the bind across different modalities because the language modality is well-explored and contains rich semantics, and freezes the language encoder acquired by VL pretraining, then train encoders for other modalities with contrastive learning.
- Recurrent Fusion Network for Image Captioning286
This paper proposes a novel recurrent fusion network (RFNet) for the image captioning task, which can exploit the interactions among the outputs of the image encoders and generate new compact and informative representations for the decoder.
- VideoMoCo: Contrastive Video Representation Learning with Temporally Adversarial Examples274
This paper improves the temporal feature representations of MoCo from two perspectives and improves MoCo temporally based on contrastive learning by empowering the temporal robustness of the encoder and modeling the temporal decay of the keys.
- Bidirectional Attentive Fusion with Context Gating for Dense Video Captioning239
This work proposes a bidirectional proposal method that effectively exploits both past and future contexts to make proposal predictions, and proposes a novel context gating mechanism to balance the contributions from the current event and its surrounding contexts dynamically.
- 236
- Temporally Grounding Language Queries in Videos by Contextual Boundary-Aware Prediction212
This work proposes an end-to-end boundary-aware model, which uses a lightweight branch to predict semantic boundaries corresponding to the given linguistic information, and outperforms its competitors with a clear margin on three public datasets.
- Controllable Video Captioning With POS Sequence Guidance Based on Gated Fusion Network189
A gating strategy is proposed to dynamically and adaptively incorporate the global syntactic POS information into the decoder for generating each word, which not only boosts the video captioning performance but also improves the diversity of the generated captions.
- Learning Modality Interaction for Temporal Sentence Localization and Event Captioning in Videos114
This work proposes a novel method for learning pairwise modality interactions in order to better exploit complementary information for each pair of modalities in videos and thus improve performances on both tasks.
- Regularizing RNNs for Caption Generation by Reconstructing the Past with the Present97
A novel architecture, namely Auto-Reconstructor Network (ARNet), which, coupling with the conventional encoder-decoder framework, works in an end-to-end fashion to generate captions, and can help regularize the transition dynamics of recurrent neural networks (RNNs).
- Learning to Guide Decoding for Image Captioning78
An extension of the encoder-decoder framework by adding a component called guiding network, which models the attribute properties of input images, and its output is leveraged to compose the input of the decoder at each time step.
- Robust dictionary learning with capped l 1 -norm78
Aiming at learning dictionaries resistant to outliers, this paper proposed capped l1-norm based dictionary learning and an efficient iterative re-weighted algorithm to solve the problem.
- VLMixer: Unpaired Vision-Language Pre-training via Cross-Modal CutMix63
A new unpaired VLP method, dubbed as VLMixer, is presented, that integrates CMC with contrastive learning to pull together the uni- modal and multi-modal views for better instance-level alignments among different modalities.
- SWEM: Towards Real-Time Video Object Segmentation with Sequential Weighted Expectation-Maximization61
A novel Sequential Weighted Expectation-Maximization (SWEM) network to greatly reduce the redundancy of memory features and maintains a fixed number of template features in memory, which ensures the stable inference complexity of the VOS system.
- Transfer Spectral Clustering56
This paper proposes a new algorithm called transfer spectral clustering (TSC), which makes use of co-clustering to achieve and control the knowledge transfer among tasks and can greatly improve the clustering performance when compared with other state-of-the-art clustering algorithms.
- 35
- Bidirectional image-sentence retrieval by local and global deep matching33
The proposed local and global deep matching model outperforms the state-of-the-art bidirectional retrieval approaches on the Flickr8K, Flickr30K, and MSCOCO datasets and the image and sentence representations exploited in local andglobal levels are demonstrated to play synergic and complementary roles for biddirectional image-sentence retrieval.
- A Deep Bayesian Tensor-Based System for Video Recommendation32
An empirical study for video recommendation demonstrates the superiority of the proposed deep Bayesian probabilistic tensor framework and indicates that it can better capture the latent patterns of interactions and generates interesting recommendations based on creative tag combinations.
- 30
- The l2,1-Norm Stacked Robust Autoencoders for Domain Adaptation30
The experimental results demonstrate that the proposed method is very effective in multiple cross domain classification datasets which include Amazon review dataset, spam dataset from ECML/PKDD discovery challenge 2006 and 20 newsgroups dataset.
- Hierarchical Photo-Scene Encoder for Album Storytelling29
The proposed model can be trained in an end-to-end manner, which results in an improved performance over the state-of-the-arts on the public visual storytelling (VIST) dataset.
- 28
- Knowledge transfer for spectral clustering26
A novel algorithm called transfer spectral clustering (TSC) is proposed, which assumes the feature embeddings being shared with the auxiliary tasks and utilizes co-clustering to extract useful information from the auxiliary datasets to improve the clustering performance.
- A trace ratio maximization approach to multiple kernel-based dimensionality reduction26
A multiple kernel learning framework for dimensionality reduction based on regularized trace ratio, termed as MKL-TR, which demonstrates its effectiveness in benchmark datasets, for supervised, unsupervised as well as semi-supervised settings.
- 21
- Learning Grounded Vision-Language Representation for Versatile Understanding in Untrimmed Videos17
A grounded vision-language learning framework for untrimmed videos, which automatically detects informative events and effectively excavates the alignments between multi-sentence descriptions and corresponding event segments, and proposes a novel semantic-aware cost to mitigate the sub-optimal matching results caused by ambiguous boundary annotations.
- In vivo evaluation of retinal ganglion cells and optic nerve's integrity in large animals by multi-modality analysis15
An optimized system to evaluate integrity of RGCs and ON in large animals in vivo is established, facilitating usage of large animal models of optic nerve diseases.
- 13
- 10
- 10
- Network Alarm Flood Pattern Mining Algorithm Based on Multi-dimensional Association8
This paper proposes the concept of association matrix pattern mining, which means that before mining the data, the multi-dimensional information of the data is used to construct the association matrices between the items and develops a conditional pattern mining algorithm based on the association matrix.
- Theoretic Analysis and Extremely Easy Algorithms for Domain Adaptive Feature Learning8
This paper provides comprehensive analysis of feature learning algorithms used in conjunction with linear classifiers for domain adaptation and proposes a novel extremely easy feature learning algorithm that is extended by leveraging multiple layers, leading to a deep linear model.
- Clinical and cortical similarities identified between bipolar disorder I and schizophrenia: A multivariate approach7
These multivariate results show that diagnostic boundaries are not clearly related to structural differences or distinct symptom profiles, and add support that BP tend to have less severe symptom profiles when compared to SZ on the PANSS without a clear distinction.
- Feature selection using autoencoders with Bayesian methods to high-dimensional data7
A hybrid approach consisted of an autoencoder and Bayesian methods is proposed for a novel feature selection and it is found that the way consisted of autoencoders and probabilistic correction methods is more meaningful than that of stacking architectures or adding constraints to autoen coders as regards feature selection.
- 7
- From Chatbot to Digital Colleague: The Paradigm Shift Toward Persistent Autonomous AI6
This work conceptualizes this transition as a shift from Chatbot to Digital Colleague: from conversational answers to persistent work, and organizes this transition along two tightly coupled dimensions.
- Federated Large Language Models: Feasibility, Robustness, Security and Future Directions6
An exhaustive survey of existing studies on FLLM feasibility is presented, methods to enhance robustness in the face of resource, data, and task heterogeneity are introduced, and novel risks associated with this integration are analyzed, including privacy threats and security challenges.
- 6
- Self-Supervised Video Action Localization with Adversarial Temporal Transforms6
This work designs a set of temporal transform operations, including naive temporal down-sampling to learnable attention-piloted time warping, and devise a self-refine module to enhance the completeness of action intervals harnessing temporal and semantic contexts.
- Evaluating clinical competencies of large language models with a general practice benchmark5
It is found that current LLMs are not suitable for autonomous deployment in clinical general practice and that all realistic applications require continuous human oversight; further optimization specifically tailored to the daily responsibilities of GPs remains essential.
- Noncontrast CT-based deep learning for predicting intracerebral hemorrhage expansion incorporating growth of intraventricular hemorrhage5
The deep learning model can identify ICH patients with high-risk rHE with favorable predictive performance than traditional baseline models based on clinical-radiologic variables and radiomics features.
- Mitigating Catastrophic Forgetting in Multi-domain Chinese Spelling Correction by Multi-stage Knowledge Transfer Framework5
A novel model-agnostic Multi-stage Knowledge Transfer (MKT) framework is proposed, which utilizes a continuously evolving teacher model for knowledge transfer in each domain, rather than focusing solely on new domain knowledge.
- EXCGEC: A Benchmark for Edit-Wise Explainable Chinese Grammatical Error Correction4
The task of EXplainable GEC (EXGEC), which focuses on the integral role of correction and explanation tasks, is introduced and a comprehensive evaluation suite is built by leveraging existing automatic metrics and conducting human evaluation experiments to demonstrate the human consistency of the automatic metrics for free-text explanations.
- Unsupervised domain adaptation via causal-contrastive learning4
This paper proposes causal invariance contrastive adaptation (CICA), a causal-contrastive learning-based unsupervised domain adaptation model for image classification that effectively separate domain-invariant features from spurious features using adversarial training, thereby reducing the interference of confounding factors on the domain adaptation task.
- 4
- Multi-Target Invisibly Trojaned Networks for Visual Recognition and Detection4
This work forms trigger embedding as an image steganography-and-steganalysis problem that conceals a secret image in another image in a decipherable and almost invisible way, and proposes a generic split-and-merge scheme for training a trojaned model.
- 3
- Construction of a diagnostic model for temporal lobe epilepsy using interpretable deep learning: disease-associated markers identification3
An interpretable deep learning diagnostic model, which combines with traditional diagnostic methods may significantly improve the accuracy of TLE diagnosis, serving as a supplementary tool for clinical assessment.
- 3
- Few-Shot Class-Incremental Learning with Prior Knowledge3
Experimental results indicate that LwPK effectively enhances the model resilience against catastrophic forgetting, with theoretical analysis based on empirical risk minimization and class distance measurement corroborating its operational principles.
- 2
- An end-to-end deep learning pipeline for hematoma expansion prediction in spontaneous intracerebral hemorrhage based on non-contrast computed tomography2
An automated pipeline for HE prediction using non-contrast computed tomography from 2020 patients across five centers is developed, suggesting that the proposed modular approach provides a promising front-line tool for rapid HE risk stratification in acute care settings.
- Causal Inference with Double/Debiased Machine Learning for Evaluating the Health Effects of Multiple Mismeasured Pollutants2
This paper used a linear regression calibration model, fitted with generalized estimating equations in an external validation study, and extended a double/debiased machine learning (DML) approach to correct for measurement error and estimate the effect of interest in the main study, demonstrating that the DML estimator with regression calibration is consistent and derived its asymptotic variance.
- Can Decentralized Stochastic Minimax Optimization Algorithms Converge Linearly for Finite-Sum Nonconvex-Nonconcave Problems?2
This work develops two novel decentralized stochastic variance-reduced gradient descent ascent algorithms for the finite-sum nonconvex-nonconcave problem that satisfies the Polyak-{\L}ojasiewicz (PL) condition and achieves linear convergence rates.
- Navigation-guided nasal endoscopic surgery for acute vision loss caused by fibrous dysplasia: a case report and review of literatures2
Transnasal endoscopic surgery guided by navigation with drainage and curettage of this bone cyst and orbital decompression resulted in progressive improvement in visual acuity that returned to normal 1 month post-operatively.
- Characteristics of post‐traumatic embitterment disorder of inpatients in a general hospital in China2
The present study shows that embitterment is a critical negative emotion in inpatients with somatic illnesses with comorbid depression, anxiety, and somatic symptoms and it is crucial to enhance PTED screening and intervention in the future.
- Fine-grained hand gesture recognition based on active acoustic signal for VR systems2
This paper leverages Ultragloves, a low cost interaction system using microphone-implanted gloves to extract the hand gestures, and designs a CNN-LSTM like learning algorithm to Extract the gestures.
- Algorithm and Theoretical Analysis for Domain Adaptation Feature Learning with Linear Classifiers.2
An formal analysis of feature learning algorithms for domain adaptation with linear classifiers shows that in order to achieve good adaptation performance, the second moments of source domain distribution and target domain distribution should be similar.
- Active Learning from Scene Embeddings for End-to-End Autonomous Driving1
This work proposes an active learning framework that relies on the BEV features in end-to-end models, called SEAD, and shows that it only needs 30\% of the nuScenes training data to achieve performance close to what can be achieved with the full dataset.
- A generative adversarial network face recognition algorithm based on mean feature matching1
It is concluded that the Generative Adversarial Network framework model based on mean feature matching proposed in this paper has higher accuracy and stronger robust, and has achieved the most advanced results in various facial benchmark tests.
- Discriminator-free adversarial domain adaptation with information balance1
A Dynamic Balance-based Domain Adaptation (DBDA) method for self-correlated domain adaptive image classification is proposed that ensures an explicit domain alignment and category distinction, thus enabling DBDA to fully leverage the predicted discriminative information for an effective feature alignment.
- A graph convolutional network face recognition algorithm based on clustering1
Experimental results demonstrate that the clustering-based GCN model achieves state-of-the-art performance on various public face datasets, highlighting the effectiveness and potential of clusteringbased GCN in face recognition tasks.
- 1
- 1
- 1
- Effects of light therapy on depression, anxiety and sleep quality in mental disorders: A systematic review and meta-analysis–
LT had a positive effect on depression, anxiety and subjective sleep quality, and LT duration, illuminance, disease subtype and medication status might be significant influencing factors.
- –
- –
- A convolutional neural network face recognition algorithm based on bootstrapping–
The proposed Bootstrapping based Convolutional Neural Network framework model can achieve Deep Learning and subsequent prediction of large-scale noisy labeled facial data and has the advantage of low computational cost.
- –
- mRNA Vaccine: The Starting Point of the Human Body Revolution–
A researcher proposes that a more plausible reason lies in +1 ribosomal frameshifting of certain tRNAs during the translation of in vitro transcribed mRNA in vivo, contributing to addressing the issue of "+1 frameshifting," enhancing the functionality of in vitro-transcribed mRNA.
- RigLSTM: Recurrent Independent Grid LSTM for Generalizable Sequence Learning–
This paper proposes recurrent independent Grid LSTM (RigLSTM), composed of a group of independent L STM cells that cooperate with each other, for exploiting the underlying modular structure of the target task.
- –
- –
- Recurrent Nested Model for Sequence Generation–
This paper proposes a module that can be plugged into the middle between the encoder and decoder to increase the depth of the whole model and follows a nested structure, which is divided into blocks with each block containing several recurrent transition steps.
- –
- –
- Learning representations with information on hand–
A multiple kernel learning framework for dimensionality reduction based on regularized trace ratio, termed as MKL-TR, is developed and aims at learning a transformation into a space of lower dimension and a corresponding kernel from the given base kernels among which some may not be suitable for the given data.
- The Application of Visual Basic on Database–
This text visits the an analyses of several kinds of technology of the database through VB, in the hope of offering some facilities for the thing that develops in the high-emcient, beautiful database application program.
- –
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.