Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Dezhong PengSuggested from co-authorship
- Suggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works72 from public data
- Contrastive Clustering862
An online clustering method called Contrastive Clustering (CC) which explicitly performs the instance- and cluster-level contrastive learning and jointly learns representations and cluster assignments in an end-to-end manner is proposed.
- 650
- 467
- 279
- 199
- 189
- 161
- Noisy-Correspondence Learning for Text-to-Image Person Re-Identification156
A novel Robust Dual Embedding method (RDE) that can learn robust visual-semantic associations even with noisy correspondence and achieves state-of-the-art results both with and without synthetic noisy correspondences on all three datasets.
- Partially View-aligned Clustering151
A novel multi-view clustering method termed partially view-aligned clustering (PVC), which proposes to use a differentiable surrogate of the non-differentiable Hungarian algorithm and recasts it as a pluggable module to establish the category-level correspondence of the unaligned data.
- 138
- Deep Evidential Learning with Noisy Correspondence for Cross-modal Retrieval136
A generalized Deep Evidential Cross-modal Learning framework (DECL), which integrates a novel Cross- modal Evidential Learning paradigm (CEL) and a Robust Dynamic Hinge loss (RDH) with positive and negative learning to address the amplification problem and embrace higher robustness against high noise.
- Scalable Deep Multimodal Learning for Cross-Modal Retrieval134
Comprehensive experimental results on four widely-used benchmark datasets demonstrate that the proposed SDML is effective and efficient in multimodal learning and outperforms the state-of-the-art methods in cross-modal retrieval.
- 127
- 120
- MaIR: A Locality-and Continuity-Preserving Mamba for Image Restoration114
A novel Mamba-based Image Restoration model (MaIR), which consists of Nested S-shaped Scanning strategy (NSS) and Sequence Shuffle Attention block (SSA), which preserves locality and continuity of the input images through the stripe-based scanning region and the S-shaped scanning path, respectively.
- Multi-View Linear Discriminant Analysis Network106
A Multi-view Linear Discriminant Analysis Network (MvLDAN) is proposed by seeking a nonlinear discriminant and view-invariant representation shared among multiple views by employing multiple feedforward neural networks and a novel eigenvalue-based multi-view objective function.
- 93
- OPQ: Compressing Deep Neural Networks with One-shot Pruning-Quantization84
This work proposes a novel One-shot Pruning-Quantization (OPQ), the first work that reveals pre-trained model is sufficient for solving pruning and quantization simultaneously, without any complex iterative/manual optimization at the finetuning stage, and proposes a unified channel-wise quantization method that leads to low bit-rate allocation without introducing extra overhead.
- Multimodal adversarial network for cross-modal retrieval74
A Multimodal Adversarial Network (MAN) method to project the multimodal data into a common space wherein the similarities between different modalities can be directly computed by the same distance measurement, and could be one of the first works to specifically design for the multi-modality representation learning with adversarial learning.
- 71
- DRSL: Deep Relational Similarity Learning for Cross-modal Retrieval69
Deep Relational Similarity Learning (DRSL) is a deep hybrid framework that integrates the relation networks module for relation learning, capturing the implicit nonlinear distance metric.
- Semantic Invariant Multi-View Clustering With Fully Incomplete Information66
This work discovers the existence of invariant semantic distribution across different views, which enables SMILE to alleviate the cross-view discrepancy to learn consensus semantics without requiring any paired samples.
- 54
- 47
- Dual Self-Paced Cross-Modal Hashing45
This work proposes a novel Dual Self-Paced Cross-Modal Hashing (DSCMH) that mimics human cognitive learning to learn hashing from ``easy'' to ``hard'' in both instance and feature levels, thereby embracing robustness against noise/outliers.
- Adversarial correlated autoencoder for unsupervised multi-view representation learning40
AdvCAE could be one of the first unsupervised multi-view representation learning approaches that work in the manner of adversarial learning and remarkably outperforms than 15 state-of-the-art methods.
- 36
- Separated Variational Hashing Networks for Cross-Modal Retrieval35
This paper proposes a novel approach called Separated Variational Hashing Networks (SVHNs), which adopts a label network (LabNet) to exploit available and nonspecific label annotations to learn a latent common Hamming space by projecting each semantic label into a common binary representation.
- 34
- Robust Variational Contrastive Learning for Partially View-unaligned Clustering34
A robust framework, dubbed VariatIonal ConTrAstive Learning (VITAL), designed to learn both common and specific information simultaneously simultaneously, which outperforms ten state-of-the-art deep clustering baselines, demonstrating its efficacy in both partially and fully aligned scenarios.
- Human-centered Interactive Learning via MLLMs for Text-to-Image Person Re-identification30
An Interactive Cross-modal Learning framework (ICL), which leverages human-centered interaction to enhance the discriminability of text queries through external multimodal knowledge, and introduces a novel Reorganization Data Augmentation (RDA) strategy based on information enrichment and diversity enhancement to enhance query discriminability.
- Robust Contrastive Multi-view Clustering against Dual Noisy Correspondence30
Contextually-spectral based correspondence refinery (CANDY) dexterously exploits inter-view similarities as context to uncover false negatives and employs a spectral-based module to denoise correspondence, alleviating the negative influence of false positives.
- Robust Contrastive Cross-modal Hashing with Noisy Labels29
This paper proposes a novel framework, termed Noise Resistance Cross-modal Hashing (NRCH), to learn hashing with noisy labels by overcoming two key challenges, i.e. noise overfitting and error accumulation.
- Semi-Supervised Multi-Modal Learning with Balanced Spectral Decomposition29
A Semi-supervised Multimodal Learning Network method (SMLN) which correlates different modalities by capturing the intrinsic structure and discriminative correlation of the multimedia data and will exactly balance all eigenvalues, thus being more competitive to existing methods.
- Deep Fair Clustering via Maximizing and Minimizing Mutual Information: Theory, Algorithm and Metric28
This work develops a mutual information theory for deep fair clustering and accordingly designs a novel algorithm, dubbed FCMI, designed to achieve four characteristics highly expected by deepFair clustering, i.e., compact, balanced, and fair clusters, as well as informative features.
- Cross-modal discriminant adversarial network28
This work proposes CDM to project the generated results into a single common space, accompanying with a novel eigenvalue-based loss, which could push as much discriminative power as possible into all latent directions.
- 28
- Robust Duality Learning for Unsupervised Visible-Infrared Person Re-Identification27
A novel robust duality learning framework (RoDE) for UVI-ReID is proposed to mitigate the adverse impact of noisy pseudo-labels and a novel Robust Adaptive Learning mechanism (RAL) is proposed to dynamically prioritize clean samples while deprioritizing noisy ones, thus avoiding overemphasizing noise.
- 27
- Local feature based multi-view discriminant analysis26
A novel local feature based multi-view discriminant analysis approach (FMDA) that addresses the small sample size (SSS) problem and preserving the discriminative information while reducing the redundant information in the LFD matrices.
- 24
- 23
- 22
- 22
- 22
- 19
- 18
- LLaVA-ReID: Selective Multi-image Questioner for Interactive Person Re-Identification17
This work proposes LLaVA-ReID, a question model that generates targeted questions based on visual and textual contexts to elicit additional details about the target person in a dialogue-based retrieval task called interactive person re-identification (Inter-ReID).
- 16
- 16
- ROAD: Robust Unsupervised Domain Adaptation with Noisy Labels14
A robust unsupervised domain adaptation framework (ROAD), which prevents the network model from overfitting noisy labels to capture accurate discrimination knowledge for domain adaptation, and a Robust Adaptive Weighted Learning mechanism (RSWL) is proposed to adaptively assign weights to each sample based on its reliability to enforce the model to focus more on reliable samples and less on unreliable samples, thereby mining robust discrimination knowledge against noisy labels in the source domain.
- 14
- Unsupervised Neural Rendering for Image Hazing13
Compared with existing haze generation methods, HazeGEN renders the hazy images in an unsupervised, learnable, and controllable manner, thus avoiding the labor-intensive efforts in paired data collection and the domain-shift issue in haze generation.
- Trustworthy Visual-Textual Retrieval12
A novel Trust-Consistent Learning framework is proposed to endow visual-textual retrieval with uncertainty evaluation for trustworthy retrieval, and first models the matching evidence according to cross-modal similarity to estimate the uncertainty for cross-modal uncertainty-aware learning.
- 11
- Learning with Noisy Correspondence10
The experimental results show the necessity of the explicit solution to the NC faced by not only the standard model training paradigm but also the pre-training and fine-tuning paradigms.
- PSRR-MaxpoolNMS: Pyramid Shifted MaxpoolNMS with Relationship Recovery10
This paper proposes a general, parallelizable and configurable approach, to completely replace GreedyNMS at all stages in all detectors, and provides a fully parallelizable solution for customized hardware design, which can be reused for accelerating NMS everywhere.
- 9
- 8
- Relationship Quantification of Image Degradations7
By quantifying degradation relationship using DRI, it is revealed that a positive DRI consistently indicates performance improvement when a beneficial auxiliary degradation is incorporated during training and the proportion of auxiliary degradation is crucial to the anchor task performance.
- 6
- 5
- 5
- PrefAce: Face-Centric Pretraining with Self-Structure Aware Distillation4
A novel self-supervised face-centric pretraining framework, called PrefAce, which learns transferable video facial representation without labels, that can transfer across various facial analysis tasks, e.g., Facial Attribute Recognition (FAR), Facial Expression Recognition (FER), DeepFake Detection (DFD), and Lip Synchronization.
- Learning Robust Multi-view Representation Using Dual-masked VAEs3
A holistic method, called Dual-masked Variational Autoencoders (DualVAE), which aims at learning robust multi-view representation and exhibits an innovative amalgamation of dual-masked prediction, mixture-of-experts learning, representation disentangling, and a joint loss function in wrapping up all components.
- 3
- 3
- Disentangling Multi-view Representations via Curriculum Learning with Learnable Prior2
A novel curriculum learning approach that adjusts the whole model to learn view-consistent representations first and then progressively view-specific representations, and incorporates a mixture-of-experts layer and a disentangling module to further enhance the quality of the learned representations.
- –
- –
- Multidimensional evaluation and scenario-adaptive application of sentiment analysis algorithms–
This study provides empirical evidence for scenario-based algorithm selection and reveals the intrinsic correlation between text feature representation and model architecture.
- Face Linear Discriminant Analysis Based on Graph Embedding and Regularization–
The unsupervised optimal class separate criterion is built, and a method to get this projection vector on the graph embedding frame is proposed that can get a huge save of computational consumption and utilize the class information of samples more effectively.
Publication data from OpenAlex; 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.