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
- Suggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works88 from public data
- Memorizing Normality to Detect Anomaly: Memory-Augmented Deep Autoencoder for Unsupervised Anomaly Detection1,828
The proposed memory-augmented autoencoder called MemAE is free of assumptions on the data type and thus general to be applied to different tasks and proves the excellent generalization and high effectiveness of the proposed MemAE.
- From Motion Blur to Motion Flow: A Deep Learning Solution for Removing Heterogeneous Motion Blur439
This work directly estimates the motion flow from the blurred image through a fully-convolutional deep neural network (FCN) and recovers the unblurred image from the estimated motion flow and is the first universal end-to-end mapping from the blur image to the dense motion flow.
- Attention-Guided Network for Ghost-Free High Dynamic Range Imaging313
The proposed AHDRNet is a non-flow-based method, which can also avoid the artifacts generated by optical-flow estimation error, and can achieve state-of-the-art quantitative and qualitative results.
- Knowledge Adaptation for Efficient Semantic Segmentation253
This work proposes a knowledge distillation method tailored for semantic segmentation to improve the performance of the compact FCNs with large overall stride and optimize the feature similarity in a transferred latent domain formulated by utilizing a pre-trained autoencoder.
- Seeing Deeply and Bidirectionally: A Deep Learning Approach for Single Image Reflection Removal191
This work argues that, to remove reflection truly well, it should estimate the reflection and utilize it to estimate the background image, and proposes a cascade deep neural network, which estimates both the Background image and the reflection.
- 181
- Learning Deep Gradient Descent Optimization for Image Deconvolution120
A recurrent gradient descent network (RGDN) is proposed by systematically incorporating deep neural networks into a fully parameterized gradient descent scheme and learns an implicit image prior and a universal update rule through recursive supervision.
- COVID-19 Chest CT Image Segmentation Network by Multi-Scale Fusion and Enhancement Operations109
A new deep convolutional neural network tailored for segmenting the chest CT images with COVID-19 infections is established and achieves state-of-the-art performance.
- 108
- RGBD Based Dimensional Decomposition Residual Network for 3D Semantic Scene Completion107
A light-weight Dimensional Decomposition Residual network (DDR) for 3D dense prediction tasks is introduced and the proposed multi-scale fusion mechanism for depth and color image can improve the completion and segmentation accuracy simultaneously.
- 85
- Part-Guided Attention Learning for Vehicle Instance Retrieval84
A Part Attention Module (PAM) is proposed to adaptively locate the most discriminative regions with high-attention weights and suppress the distraction of irrelevant parts with relatively low weights and achieves new state-of-the-art vehicle IR performance on four large-scale benchmark datasets.
- 84
- Self-Expansion of Pre-trained Models with Mixture of Adapters for Continual Learning73
Self-Expansion of pre-trained models with Modularized Adaptation (SEMA), a novel approach to enhance the control of stability-plasticity balance in PTM-based CL, enables better knowledge reuse and sub-linear expansion rate.
- Learning with Mixture of Prototypes for Out-of-Distribution Detection71
PrototypicAl Learning with a Mixture of prototypes (PALM) is proposed which models each class with multiple prototypes to capture the sample diversities, and learns more faithful and compact samples embeddings to enhance OOD detection.
- Continual All-in-One Adverse Weather Removal With Knowledge Replay on a Unified Network Structure59
A novel continual learning framework with effective knowledge replay (KR) on a unified network structure is developed, which performs competitively to existing dedicated or joint training image restoration methods.
- Learning Bayesian Sparse Networks with Full Experience Replay for Continual Learning56
A Sparse neural Network for Continual Learning (SNCL), which employs variational Bayesian sparsity priors on the activations of the neurons in all layers and achieves state-of-the-art result for mitigating forgetting.
- Learning to Fuse Monocular and Multi-view Cues for Multi-frame Depth Estimation in Dynamic Scenes53
A novel method to learn to fuse the multi-view and monocular cues encoded as volumes without needing the heuristically crafted masks is proposed, which includes the cross-cue attention (CCA) to encode the spatially non-local relative intra-relations from each source to enhance the representation of the other.
- Dual-Attention-Guided Network for Ghost-Free High Dynamic Range Imaging51
This work proposes a novel dual-attention-guided end-to-end deep neural network, called DAHDRNet, which produces high-quality ghost-free HDR images and employs a hybrid loss function, which consists of a perceptual loss, a total variation loss, and a content loss to recover photo-realistic images.
- Learning and Memorizing Representative Prototypes for 3D Point Cloud Semantic and Instance Segmentation48
This paper proposes a memory-augmented network to learn and memorize the representative prototypes that cover diverse samples universally, and introduces a memory module to alleviate the forgetting issue by recording the patterns seen in mini-batch training.
- 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.
- 41
- Part-Guided Attention Learning for Vehicle Re-Identification.39
A Part Attention Module (PAM) is proposed to adaptively locate the most discriminative regions with high-attention weights and suppress the distraction of irrelevant parts with relatively low weights and achieves new state-of-the-art vehicle Re-ID performance on four large-scale benchmark datasets.
- High dynamic range imaging via gradient-aware context aggregation network38
The proposed HDR imaging approach that aggregates the information from multiple LDR images with guidance from image gradient domain generates artifact-free images by integrating the image gradient information and the image context information in the pixel domain.
- DyMO: Training-Free Diffusion Model Alignment with Dynamic Multi-Objective Scheduling35
A plug-and-play training-free alignment method, DyMO, for aligning the generated images and human preferences during inference, with dynamic scheduling of multiple objectives and intermediate recurrent steps to reflect the requirements at different steps.
- Learning to Zoom-In via Learning to Zoom-Out: Real-World Super-Resolution by Generating and Adapting Degradation35
A framework to learn SR from an arbitrary set of unpaired LR and HR images and to see how far a step can go in such a realistic and “unsupervised” setting is proposed.
- 32
- Take Only What You Need: Rank Minimization as an Implicit Forgetting Regularizer in Continual Learning31
A universal and efficient Continual Learning approach for VLM based on Dynamic Rank-Selective LoRA (CoDyRA), which directly improves the PTMs while preserving the existing knowledge from both pre-training and CL, and preserves the original model architecture and deployment pipeline.
- The Neural Process Family: Survey, Applications and Perspectives31
A comprehensive survey of NPF models is needed to organize and relate their motivation, methodology, and experiments and shed light on their potential to bring several recent advances in other deep learning domains under one umbrella.
- Deblurring Natural Image Using Super-Gaussian Fields31
A novel image prior for image deblurring based on a Super-Gaussian field model with adaptive structures is presented, which captures the complicated structures in natural images by integrating potentials on all cliques into a joint probabilistic distribution.
- End-To-End Diagnosis And Segmentation Learning From Cardiac Magnetic Resonance Imaging30
This paper proposes a learning method to train diagnosis models based on multi-task learning that jointly trains for the tasks of segmentation and diagnosis classification, and hypothesizes that segmentation has a regularizing effect on the learning of features relevant for diagnosis.
- 28
- Identifiable Latent Polynomial Causal Models Through the Lens of Change27
The scope of latent causal models is extended to involve nonlinear causal relationships, represented by polynomial models, and general noise distributions conforming to the exponential family, and a novel empirical estimation method is proposed, grounded in the theoretical finding, that enables learning consistent latent causal representations.
- 20
- Variational Bayesian Dropout With a Hierarchical Prior20
This work presents a new generalization of Gaussian dropout, termed variational Bayesian dropout (VBD), which turns to exploit a hierarchical prior on the network weights and infer a new joint posterior, and implements the hierarchical prior as a zero-mean Gaussian distribution with variance sampled from a uniform hyper-prior.
- Learning an Optimizer for Image Deconvolution.20
A Recurrent Gradient Descent Network (RGDN) is proposed by systematically incorporating deep neural networks into a fully parameterized gradient descent scheme and learns an implicit image prior and a universal update rule through recursive supervision.
- I Predict Therefore I Am: Is Next Token Prediction Enough to Learn Human-Interpretable Concepts from Data?17
A novel generative model that generates tokens on the basis of concepts formulated as latent discrete variables is introduced that provides evidence that LLMs capture essential underlying generative factors, offers a unified and principled perspective for understanding the linear representation hypothesis, and motivates a theoretically grounded approach for evaluating sparse autoencoders.
- 16
- 16
- ARSAC: Efficient model estimation via adaptively ranked sample consensus15
A novel Adaptively Ranked Sample Consensus (ARSAC) algorithm is presented to boost the speed and robustness of RANSAC, and adopts non-uniform sampling based on the ranked measurements to speed up the sampling process.
- MPGL: An Efficient Matching Pursuit Method for Generalized LASSO15
This work formulate the GL problem as a convex quadratic constrained linear programming (QCLP) problem and tailor-make a cutting plane method, and proposes an efficient Matching Pursuit Generalized LASSO (MPGL) method, which overcomes issues, and is guaranteed to converge to a global optimum.
- Maximizing Spatio-Temporal Entropy of Deep 3D CNNs for Efficient Video Recognition14
This work proposes to automatically design efficient 3D CNN architectures via a novel training-free neural architecture search approach tailored for3D CNNs considering the model complexity, and proposes a spatio-temporal entropy score (STEntr-Score) with a refinement factor to handle the discrepancy of visual information in spatial and temporal dimensions.
- 14
- 14
- Boundarymix: Generating pseudo-training images for improving segmentation with scribble annotations13
BoundaryMix, a simple-but-effective solution, which generates pseudo training image-annotation pairs from the original images to supplement the missing semantic boundaries and improves the overall segmentation performance.
- 12
- EGGen: Image Generation with Multi-entity Prior Learning through Entity Guidance11
This work advances the stable diffusion model by segmenting comprehensive prompts into distinct entity-specific prompts with bounding boxes, enabling a transition from multi-entity to single-entity generation in cross-attention layers and introducing entity-centric cross-attention layers that focus on individual entities to preserve their uniqueness and accuracy.
- Improved genomic prediction using machine learning with Variational Bayesian sparsity11
This research explores the use of an ML architecture governed by variational Bayesian sparsity in its initial layers that is called VBS-ML, a computationally feasible approach for improving genomic prediction conducted with large breeding population numbers and genetic markers.
- 11
- Blind image deblurring by promoting group sparsity10
The proposed group sparsity priors are based on the fact that the nonzero elements of natural image gradients and blur kernels tend to cluster in structured group pattern and are introduced, which preserve the salient structures and smooth the minor components in imageGradients and restrict the blur kernel in a domain of dynamic group sparse vector.
- 9
- 9
- 8
- Continual Learning on CLIP via Incremental Prompt Tuning with Intrinsic Textual Anchors7
This paper proposes a concise CL approach for CLIP based on incremental prompt tuning that fully exploits its multi-modal structure and the stability of textual representations, and introduces a relational diversity regularization on the textual anchors to prevent embedding space collapse and mitigate correlated forgetting.
- Beyond DAGs: A Latent Partial Causal Model for Multimodal Learning7
This work introduces a unified causal model specifically designed for multimodal data, and shows that multimodal contrastive representation learning excels at identifying latent coupled variables within the proposed unified model, up to linear or permutation transformations resulting from different assumptions.
- Premonition: Using Generative Models to Preempt Future Data Changes in Continual Learning7
It is shown here that the combination of a large language model and an image generation model can similarly provide useful premonitions as to how a continual learning challenge might develop over time, thus becoming a valuable input to any existing continual learning method.
- FocDepthFormer: Transformer with Latent LSTM for Depth Estimation from Focal Stack7
A novel Transformer-based network, FocDepthFormer, which integrates a Transformer with an LSTM module and a CNN decoder and outperforms state-of-the-art approaches across multiple evaluation metrics.
- SDGE: Stereo Guided Depth Estimation for 360°Camera Sets5
The Stereo Guided Depth Estimation method is proposed, which enhances depth estimation of the full image by explicitly utilizing multi-view stereo results on the overlap by employing a self-calibration method to obtain highly accurate relative poses of the adjacent cameras with minor overlap.
- Learning Mamba as a Continual Learner: Meta-Learning Selective State Space Models for Continual Learning5
This paper proposes MambaCL, a meta-learned continual learner, a meta-learned sequence-prediction-based continual learners without retaining all past representations, and introduces selectivity regularization to enhance MambaCL’s training.
- UIR-LoRA: Achieving Universal Image Restoration through Multiple Low-Rank Adaptation5
This work proposes a universal image restoration framework based on multiple low-rank adapters (LoRA) from multi-domain transfer learning that leverages the pre-trained generative model as the shared component for multi-degradation restoration and transfers it to specific degradation image restoration tasks using low-rank adaptation.
- Identifying Weight-Variant Latent Causal Models5
This work analysis three intrinsic properties in latent space, including transitivity, permutation and scaling, and proposes a novel method, termed Structural caUsAl Variational autoEncoder (SuaVE), which directly learns latent causal variables, together with the mapping from the latent causal variable to the observed ones.
- Semi-supervised Learning via Conditional Rotation Angle Estimation5
This work proposes a new SemSL approach called Conditional Rotation Angle EStimation (CRAES), which is featured by adopting a module which predicts the image rotation angle conditioned on the candidate image class and achieves superior performance over the other existing ways of combining SlfSL and SemSL.
- 5
- Hierarchical Point-Patch Fusion with Adaptive Patch Codebook for 3D Shape Anomaly Detection4
A hierarchical point-patch anomaly scoring network that jointly models regional part features and local point features for robust anomaly reasoning and an adaptive patchification module integrates self-supervised decomposition to capture complex structural deviations is proposed.
- Learning Informative Latent Representation for Quantum State Tomography4
A transformer-based autoencoder architecture tailored for QST with imperfect measurement data is proposed, which leverages a transformer-based encoder to extract an informative latent representation (ILR) from imperfect measurement data and employs a decoder to predict the quantum states based on the ILR.
- Let Your Video Listen to Your Music! -- Beat-Aligned, Content-Preserving Video Editing with Arbitrary Music4
A novel and efficient framework-termed MVAA (Music-Video Auto-Alignment)-that automatically edits video to align with the rhythm of a given music track while preserving the original visual content while preserving the original visual content is proposed.
- 4
- Mining Your Own Secrets: Diffusion Classifier Scores for Continual Personalization of Text-to-Image Diffusion Models4
Inspired by the successful CL methods that rely on class-specific information for regularization, the inherent class-conditioned density estimates, also known as diffusion classifier (DC) scores, are resorted to for continual personalization of text-to-image diffusion models.
- Image editing with varying intensities of processing4
The hypothesis is that the part of parameters that correspond to the intensity of processing of the network are implicitly embedded in the whole parameter set and can be explicitly extracted and utilized, even with limited supervision, in image editing with Varying Intensities of Processing.
- On Token's Dilemma: Dynamic MoE with Drift-Aware Token Assignment for Continual Learning of Large Vision Language Models3
A token-level assignment guidance steers ambiguous and old tokens away from new experts to preserve established routing patterns and alleviate routing-drift-induced forgetting, while complementary routing score regularizations enforce expert-group separation and promote new-expert specialization.
- Balancing Plasticity and Stability via Dual-Branch Learning in Online Continual Learning3
A novel dual-branch framework designed to manage the plasticity-stability trade-off in CL is proposed, which integrates new information while maintaining robust retention of prior knowledge.
- Latent Covariate Shift: Unlocking Partial Identifiability for Multi-Source Domain Adaptation3
It is shown that although (completely) identifying the proposed latent causal model is challenging, the latent content variable can be identified up to scaling by using its dependence with labels from source domains, together with the identifiability conditions of nonlinear ICA.
- 3
- A Patch-Based Non-local Means Method for Image Denoising3
A revised version of non-local means denoising method using image patches to implement non- Local Means Denoising, where detail, texture and structure information can be preserved due to the redundancy in similar image patches.
- MultiEdit: Advancing Instruction-based Image Editing on Diverse and Challenging Tasks2
This work introduces MultiEdit, a comprehensive dataset featuring over 107K high-quality image editing samples and employs a novel dataset construction pipeline that utilizes two multi-modal large language models (MLLMs) to generate visual-adaptive editing instructions and produce high-fidelity edited images, respectively.
- Little by Little: Continual Learning via Incremental Mixture of Rank-1 Associative Memory Experts2
MoRA is proposed, a daptive learning approach with self-activated and sparse rank activation with significant effectiveness on enhancing CL with PTMs, and improving generalization while mitigating forgetting.
- A Comprehensive CT Dataset for Liver Computer Assisted Diagnosis2
This work constructs the first comprehensive liver dataset, called ComLiver, which consists of multiple liver-related tasks and manually marks the elaborate labels for each task and introduces vessel instance segmentation, couinaud segmentation tasks based on previous tasks.
- Variational Bayesian Dropout.2
A new generalization of Gaussian dropout is presented, termed variational Bayesian dropout (VBD), which turns to exploit a hierarchical prior on the network weights and infer a new joint posterior, which is a proper prior which enables the inference of posterior to be well-posed.
- Learning scene-aware image priors with high-order Markov random fields2
This paper presents to learn a scene-aware image prior based on the high-order Markov random field (MRF) model (SA-MRF), which jointly learns a set of shared low-level features and different potentials for specific scene contents.
- 2
- Seeing the Unseen: Composing Outliers for Compositional Zero-Shot Learning1
A novel compositional zero-shot learning method called COMO, which composes outliers in training for distinguishing seen and unseen compositions and further applying specific strategies for them, which is effective in both the closed-world setting and the open-world setting.
- 1
- 1
- When Distillation Breaks Motion Control: Restoring Generative Trajectories for Fast Video Generators–
- –
- –
- Highway Graph to Accelerate Reinforcement Learning–
The highway graph is introduced, where edges can encapsulate multiple state transitions, enabling value propagation across multiple time steps in a single iteration, and the training process is significantly accelerated, particularly in the early stages of training.
- –
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 2026-10-14. 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.