Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks111 from public data
- Deep Dense Multi-Scale Network for Snow Removal Using Semantic and Depth Priors370
This paper proposes a Deep Dense Multi-Scale Network for snow removal by exploiting semantic and geometric priors, and designs a DDMSNet to learn semantic-aware and geometry-aware representation via a self-attention mechanism to produce the final clean images.
- 135
- 127
- On Estimating Air Pollution from Photos Using Convolutional Neural Network114
An effective convolutional neural network is devised to estimate air's quality based on photos to alleviate the vanishing gradient issue effectively and a modified activation function is developed for photo based air pollution estimation.
- Sparse representation for robust abnormality detection in crowded scenes111
This paper proposes a fully unsupervised non-negative sparse coding based approach for abnormality event detection in crowded scenes, which is specifically tailored to cope with feature noisy and uncertainty.
- Self-Supervised Information Bottleneck for Deep Multi-View Subspace Clustering104
A new framework called Self-supervised Information Bottleneck based Multi-view Subspace Clustering (SIB-MSC) is established, which can learn a latent space for each view to capture common information among the latent representations of different views by removing superfluous information from the view itself while retaining sufficient information for the latent representation of other views.
- On modeling and predicting individual paper citation count over time97
This work proposes a method for predicting the citation counts of individual publications, over an arbitrary time period, and explores paper-specific covariates and a point process model to account for the aging effect and triggering role of recent citations, through which papers lose and gain their popularity, respectively.
- 94
- Residual Invertible Spatio-Temporal Network for Video Super-Resolution93
A novel end-to-end architecture called Residual Invertible Spatio-Temporal Network (RISTN) for video super-resolution that can sufficiently exploit the spatial information from low-resolution to high-resolution, and effectively models the temporal consistency from consecutive video frames.
- Self-Paced Multi-Task Learning91
This paper proposes a novel multi-task learning framework that attempts to learn the tasks by simultaneously taking into consideration the complexities of both tasks and instances per task, by presenting a new task-oriented regularizer that can jointly prioritize tasks and instance.
- Beyond Monocular Deraining: Parallel Stereo Deraining Network Via Semantic Prior78
A Paired Rain Removal Network (PRRNet), which exploits both stereo images and semantic information is presented, and a Semantic-Aware Deraining Module (SADM) is developed which solves both tasks of semantic segmentation and deraining of scenes.
- On Scalability of Association-rule-based Recommendation71
A distributed framework that can scale different association-rule-based recommendation methods in a unified way is presented, where a generic tree-type structure is defined to store separate kinds of patterns, and an efficient algorithm is designed for mining eligible patterns along with computing recommendation scores.
- Autoencoder Regularized Network For Driving Style Representation Learning71
Experiments show that ARNet can learn a good generalized driving style representation and significantly outperforms existing methods and alternative architectures by reaching the least estimation error and the highest identification accuracy compared with traditional supervised learning methods.
- Learning Transferable Self-Attentive Representations for Action Recognition in Untrimmed Videos with Weak Supervision68
A novel weakly supervised framework to simultaneously locate action frames as well as recognize actions in untrimmed videos, which takes advantage of the self-attention mechanism to weight each frame, such that the influence of background frames can be effectively eliminated.
- AdapNet: Adaptability Decomposing Encoder–Decoder Network for Weakly Supervised Action Recognition and Localization60
A novel adaptability decomposing encoder–decoder network is proposed to transfer reliable knowledge between the trimmed and untrimmed videos for action recognition and localization by bidirectional point process modeling, given only video-level annotations.
- FreeKD55
This paper proposes the first Free-direction Knowledge Distillation framework via Reinforcement learning for GNNs, called FreeKD, which is no longer required to provide a deeper well-optimized teacher GNN, and has comparable or even better performance than traditional KD algorithms that distill knowledge from a deeper and stronger teacher Gnn.
- 55
- 46
- Detecting Adversarial Data by Probing Multiple Perturbations Using Expected Perturbation Score45
This paper proposes a new statistic called expected perturbation score (EPS), which is essentially the expected score of a sample after various perturbations, and theoretically proves that EPS is a proper statistic to compute the discrepancy between two samples under mild conditions.
- 45
- Improving Maximum Likelihood Estimation of Temporal Point Process via Discriminative and Adversarial Learning44
This paper aims to improve MLE by discriminative and adversarial learning by devising a gradient based learning procedure with two complementary recipes that directly reflects the prediction accuracy of the model and can add sharpness to the smooth effect inherently caused by the MSE loss.
- A Self-Paced Regularization Framework for Multilabel Learning42
This work proposes a new multilabel learning formulation by introducing a self-paced function as a regularizer, so as to simultaneously prioritize label learning tasks and instances in each iteration, and provides a general way to find the desiredSelf-paced functions.
- Multi-Instance Multi-Label Action Recognition and Localization Based on Spatio-Temporal Pre-Trimming for Untrimmed Videos39
A novel multi-instance multi-label modeling network based on spatio-temporal pre-trimming to recognize actions and locate corresponding frames in untrimmed videos and demonstrates the efficacy of this method when compared with the state-of-the-arts.
- Reconstruction Regularized Deep Metric Learning for Multi-Label Image Classification.39
A novel deep metric learning method to tackle the multi-label image classification problem by attempting to explore a latent space, where images and labels are embedded via two unique deep neural networks, respectively, to capture the relationships between image features and labels.
- Similarity Learning via Kernel Preserving Embedding39
It is argued that it is beneficial to preserve the overall relations when the authors extract similarity information and proposed kernel preserving opens up a large number of possibilities to embed high-dimensional data into low-dimensional space.
- Robust Knowledge Adaptation for Dynamic Graph Neural Networks34
The proposed Ada-DyGNN is the first attempt to explore robust knowledge adaptation via reinforcement learning specifically tailored for dynamic graph neural networks and achieves the state-of-the-art performance.
- Keypoint-based Progressive Chain-of-Thought Distillation for LLMs34
A token weighting module utilizing mask learning to encourage accurate mimicry of keypoint tokens by the student during distillation is proposed and an in-rationale progressive distillation strategy is developed, starting with training the student to generate the final reasoning steps and gradually extending to cover the entire rationale.
- Ensemble classifier based on optimized extreme learning machine for motor imagery classification34
The proposed extreme learning machine (ELM) based method achieved significant higher classification accuracies than those of the competing methods on both two-class and four-class motor imagery data and can be a promising candidate for accurate classification of motor imagery EEG in BCI systems.
- 31
- Towards Open Temporal Graph Neural Networks30
This paper proposes a general and principled learning approach for open temporal graphs, called OTGNet, and explores a new message passing mechanism by extending the information bottleneck principle to only propagate class-agnostic knowledge between nodes of different classes, avoiding aggregating conflictive information.
- Dynamic Graph Modeling for Weakly-Supervised Temporal Action Localization30
This paper proposes a dynamic graph-based method, namely DGCNN, to explore the two-stream relation between action segments, and demonstrates that this method is superior to the state-of-the-arts.
- On Deep Unsupervised Active Learning30
A novel Deep neural network framework for Unsupervised Active Learning, called DUAL, which can explicitly learn a nonlinear embedding to map each input into a latent space through an encoder-decoder architecture, and introduce a selection block to select representative samples in the the learnt latent space.
- 29
- End-to-end learning for image-based air quality level estimation29
An effective convolutional neural network to estimate air quality based on images by designing an ensemble CNN for air quality estimation which is expected to obtain more accurate and stable results than a single classifier.
- Modeling contagious merger and acquisition via point processes with a profile regression prior28
A tailored model learning algorithm is devised that incorporates both static profile covariates and past M&A activities, which has been well-recognized by the economics and management communities.
- 21
- Unsupervised Active Learning via Subspace Learning21
This paper proposes a novel unsupervised Active Learning model via Subspace Learning, called ALSL, which aims to discovery the low-rank structures of data, and then perform sample selection based on learnt low- rank representations.
- 20
- Semi-supervised Active Learning for Instance Segmentation via Scoring Predictions19
This paper presents an uncertainty sampling strategy named Triplet Scoring Predictions (TSP) to explicitly incorporate samples ranking clues from classes, bounding boxes and masks and devise a progressive pseudo labeling regime using the above TSP in semi-supervised manner.
- Contour detection via stacking random forest learning18
The proposed contour detection algorithm is evaluated against leading methods in the literature on the Berkeley segmentation and Weizmann horse data sets and performs favorably against state-of-the-art methods in terms of speed and accuracy.
- LRSC: Learning Representations for Subspace Clustering17
This paper aims to leverage external data through constructing lots of relevant tasks to guide the training of the encoder, motivated by the idea of meta-learning, and proposes a novel subspace clustering framework through learning precise sample representations.
- LaKD: Length-agnostic Knowledge Distillation for Trajectory Prediction with Any Length Observations16
This paper proposes a dynamic length-agnostic knowledge distillation framework, named LaKD, which can make accurate trajectory predictions, regardless of the length of observed data, and designs a dynamic soft-masking mechanism, where it first calculates the importance of neuron units and then applies soft-masking to them, to safeguard critical units from disruption during the knowledge distillation process.
- 16
- Context-aware local abnormality detection in crowded scene16
A novel algorithm by jointly modeling motion and context information targeting at detecting abnormal events in crowded scenes showing superior performance to many related state-of-the-art methods is proposed.
- 15
- 15
- 15
- Action Shuffling for Weakly Supervised Temporal Localization14
This paper analyzes the order-sensitive and location-insensitive properties of actions, and embodies them into a self-augmented learning framework to improve the weakly supervised action localization performance.
- 14
- 13
- 12
- Multiprior Learning Via Neural Architecture Search for Blind Face Restoration12
A face restoration searching network to adaptively search the suitable feature extraction architecture within the authors' specified search space, which can directly contribute to the restoration quality and the multiple facial prior searching network (MFPSNet) with a multiprior learning scheme.
- Max-Margin-Based Discriminative Feature Learning12
A new max-margin-based discriminative feature learning method is proposed, which aims at learning a low-dimensional feature representation so as to maximize the global margin of the data and make the samples from the same class as close as possible.
- ITPNet: Towards Instantaneous Trajectory Prediction for Autonomous Driving11
This work proposes a backward forecasting mechanism to reversely predict the latent feature representations of unobserved historical trajectories of the agent based on its two observed locations and then leverage them as complementary information for future trajectory prediction.
- 11
- 10
- Deep Unsupervised Active Learning on Learnable Graphs9
A novel deep unsupervised active learning model via learnable graphs, named ALLGs, which benefits from learning optimal graph structures to acquire better sample representation and select representative samples and incorporates shortcut connections among different layers.
- 9
- MORES: Online Incremental Multiple-Output Regression for Data Streams.9
A novel online multiple-output regression method, called MORES, for streaming data, that can learn the structure of the regression coefficients to facilitate the model's continuous refinement and introduce a forgetting factor to set different weights on samples so as to track the data streams' evolving characteristics quickly from the latest samples.
- TwinNet: Twin Structured Knowledge Transfer Network for Weakly Supervised Action Localization8
A novel weakly supervised framework to recognize actions and locate the corresponding frames in untrimmed videos simultaneously and takes advantage of the self-attention mechanism to obtain a compact video representation, such that the influence of background frames can be effectively eliminated.
- 7
- 7
- Shared Growth of Graph Neural Networks via Prompted Free-Direction Knowledge Distillation6
This paper proposes the first Free-direction Knowledge Distillation framework via reinforcement learning for GNNs, called FreeKD, which is no longer required to provide a deeper well-optimized teacher GNN, and develops FreeKD++ and FreeKD-Prompt to enable free-direction knowledge transfer among multiple shallow GNNs.
- Active Sample Learning and Feature Selection: A Unified Approach.6
This paper presents a unified framework to conduct active learning and feature selection simultaneously in a fully unsupervised setting by solving the NP-hardness of the raw problem efficiently by an iterative algorithm, and proves its global convergence.
- Functional magnetic resonance imaging classification based on random forest algorithm in Alzheimer's disease5
Functional connection characteristics can be effectively analyzed by the random forest algorithm which can distinguish AD, MCI and NC accurately and can provide an objective reference for the early clinical diagnosis of AD.
- 5
- Spatially Regularized Streaming Sensor Selection4
This paper proposes to perform sensor selection in a multi-variate interpolation framework, such that the data sampled by the selected sensors can well predict those of the inactive sensors, and incorporates sensors' spatial information as two regularizers, which leads to significantly better prediction performance.
- 4
- PSA-MF: Personality-Sentiment Aligned Multi-Level Fusion for Multimodal Sentiment Analysis3
A personality-sentiment aligned multi-level fusion framework is proposed and a novel personality-sentiment alignment method is proposed to obtain personalized sentiment embeddings from the textual modality for the first time.
- Importance-aware Shared Parameter Subspace Learning for Domain Incremental Learning3
This work proposes an importance-aware shared parameter subspace learning for domain incremental learning, on the basis of low-rank adaption (LoRA), and designs an importance-aware mechanism that adaptively assigns an importance weight to the domain-shared subspace for the corresponding domain.
- 3
- 3
- 3
- 3
- 2
- DREAM: Domain-Agnostic Reverse Engineering Attributes of Black-Box Model2
A general and principled framework DREAM is put forward, by casting this problem of black-box reverse engineering as out-of-distribution (OOD) generalization, which can learn a domain-agnostic meta-model to infer the attributes of the target black-box model with unknown training data.
- 2
- Weakly-Supervised Action Recognition and Localization via Knowledge Transfer2
A novel weakly-supervised action recognition framework for untrimmed videos to use only video-level annotations to transfer information from publicly available trimmed videos to assist in model learning, namely KTUntrimmedNet.
- 2
- 2
- 2
- 1
- 1
- 1
- Learning to Generate Parameters of ConvNets for Unseen Image Data1
A new hypernetwork-based model, called PudNet, which intends to learn a mapping between datasets and their corresponding network parameters, then predicts parameters for unseen data with only a single forward propagation, and benefits from a series of adaptive hyper-recurrent units sharing weights to capture the dependencies of parameters among different network layers.
- 1
- Maximizing Multi-scale Spatial Statistical Discrepancy1
Experiments on a wide range of datasets demonstrate the advantages of RefineScan over the state-of-the-art algorithms: It always finds the largest discrepancy scores and remarkably better characterizes multi-scale discrepancy boundaries.
- 1
- Collaborative Tracking: Dynamically Fusing Short-Term Trackers and Long-Term Detector1
This paper proposes a novel collaborative tracking framework fusing short-term trackers and long-term object detector, which would consist of a frame-to-frame tracker and a weakly supervised tracker which would be updated under theWeakly supervised information and re-initialized by long- term detector while the trackers fail.
- 1
- 1
- 1
- –
- –
- RL-ISLAP: A Reinforcement Learning Framework for Industrial-Scale Linear Assignment Problems at Alipay–
The proposed RL-ISLAP is an efficient and lightweight Reinforcement Learning framework for Industrial-Scale Linear Assignment Problems that achieves solutions of comparable quality in 2/3 of the time when compared to the SOTA distributed solving system employing fine-tuned empirical parameter rules.
- DREAM: Domain-free Reverse Engineering Attributes of Black-box Model–
This paper investigates a new problem of Domain-agnostic Reverse Engineering the Attributes of a black-box target Model without requiring the availability of the target model's training dataset, and puts forward a general and principled framework by casting this problem as an out of distribution (OOD) generalization problem.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Attribute Reduction Algorithm Based on Discernible Boolean Matrix and Classification Coefficient–
Two definitions, which are called respectively discernible Boolean matrix and classification coefficient, are given for increasing visual expression of some concepts and operations in the rough sets theory.
- –
- –
- A New Method of Simplifying Grammar——tree algorithm–
A new tree algorithm is introduced for deleting superfluous symbol and production in grammar , and a practical example to check the algorithm is given.
- –
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.