Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks515 from public data
- Learning Efficient Convolutional Networks through Network Slimming2,909
The approach is called network slimming, which takes wide and large networks as input models, but during training insignificant channels are automatically identified and pruned afterwards, yielding thin and compact models with comparable accuracy.
- 1,141
- 913
- Learning a Mahalanobis distance metric for data clustering and classification637
This paper considers a general problem of learning from pairwise constraints in the form of must-links and cannot-links, and aims to learn a Mahalanobis distance metric.
- 473
- 421
- An in-field automatic wheat disease diagnosis system394
Experimental results demonstrate that the proposed system outperforms conventional CNN architectures on recognition accuracy under the same amount of parameters, meanwhile maintaining accurate localization for corresponding disease areas.
- Manifold-ranking based image retrieval389
MRBIR first makes use of a manifold ranking algorithm to explore the relationship among all the data points in the feature space, and then measures relevance between the query and all the images in the database accordingly, which is different from traditional similarity metrics based on pair-wise distance.
- 384
- Trace ratio criterion for feature selection378
A novel algorithm is proposed to efficiently find the global optimal feature subset such that the subset-level score is maximized, and extensive experiments demonstrate the effectiveness of the proposed algorithm in comparison with the traditional methods for feature selection.
- 338
- 326
- 325
- Robust multi-task feature learning315
This paper proposes a Robust Multi-Task Feature Learning algorithm (rMTFL) which simultaneously captures a common set of features among relevant tasks and identifies outlier tasks, and provides a detailed theoretical analysis on the proposed rMTFL formulation.
- 311
- 287
- 272
- DeepFish: Accurate underwater live fish recognition with a deep architecture258
A framework to recognize fish from videos captured by underwater cameras deployed in the ocean observation network is proposed, using a deep architecture to extract features of the foreground fish images and a linear SVM classifier for classification.
- Learning and Innovative Elements of Strategy Adoption Rules Expand Cooperative Network Topologies223
The results suggest that a balanced duo of learning and innovation may help to preserve cooperation during the re-organization of real-world networks, and may play a prominent role in the evolution of self-organizing, complex systems.
- Semi-supervised Multi-label Learning by Solving a Sylvester Equation214
It is shown that the labels of unlabeled data finally can be obtained by solving a Sylvester Equation, and a regularization framework combining two regularization terms for the two graphs is suggested.
- 207
- Semi-Supervised Clustering via Matrix Factorization194
A novel matrix factorization based approach for semi-supervised clustering, which is extended to co-cluster the data sets of different types with constraints and shows the superiority of the proposed method.
- Current trends in the development of intelligent unmanned autonomous systems182
The trends in the development of intelligent unmanned autonomous systems are introduced by summarizing the main achievements in each technological platform by classify the relevant technologies into seven areas.
- 182
- 176
- A unified framework for semi-supervised dimensionality reduction173
This paper proposes a semi-supervised dimensionality reduction framework, which can efficiently handle the unlabeled data and can significantly improve the accuracy rates of the corresponding supervised and unsupervised approaches.
- Weakly Supervised Contrastive Learning168
A weakly supervised contrastive learning framework (WCL) based on two projection heads, one of which will perform the regular instance discrimination task, and the other head will use a graph-based method to explore similar samples and generate a weak label to pull the similar images closer.
- 165
- 164
- Multi-Stage Multi-Task Feature Learning.163
A non-convex formulation for multi-task sparse feature learning based on a novel regularizer is proposed and a detailed theoretical analysis is presented showing that MSMTFL achieves a better parameter estimation error bound than the convex formulation.
- Multiple Kernel Clustering162
A multiple kernel clustering (MKC) algorithm that simultaneously finds the maximum margin hyperplane, the best cluster labeling, and the optimal kernel is proposed.
- GreedyNAS: Towards Fast One-Shot NAS With Greedy Supernet159
This paper proposes a multi-path sampling strategy with rejection, and greedily filter the weak paths to ease the burden of supernet by encouraging it to focus more on evaluation of those potentially-good ones, which are identified using a surrogate portion of validation data.
- 159
- Semi-supervised orthogonal discriminant analysis via label propagation154
This paper proposes a novel semi-supervised orthogonal discriminant analysis via label propagation that propagates the label information from the labeled data to the unlabeled data through a specially designed label propagation, and thus the distribution of the unl labeled data can be explored more effectively to learn a better subspace.
- Agree to Disagree: Adaptive Ensemble Knowledge Distillation in Gradient Space144
This paper examines the diversity of teacher models in the gradient space and regard the ensemble knowledge distillation as a multi-objective optimization problem so that the method can be seen as a dynamic weighting method for each teacher in the ensemble.
- 143
- Classification of Digital Photos Taken by Photographers or Home Users142
This paper addresses a specific image classification task, i.e. to group images according to whether they were taken by photographers or home users, and shows an application in No-Reference holistic quality assessment as a natural extension of such image classification.
- Graph-based semi-supervised learning137
First, a method called linear neighborhood propagation is proposed, which can automatically construct the optimal graph, and a novel multilevel scheme is introduced to make the algorithm scalable for large data sets.
- 135
- A general graph-based semi-supervised learning with novel class discovery135
The core idea of the algorithm is to not only achieve the goal of semi-supervised learning, but also to discover the latent novel class in the data, which may be unlabeled by the user.
- An experimental evaluation of ensemble methods for EEG signal classification134
The present study systematically evaluates the performance of the three ensemble methods for EEG signal classification of mental imagery tasks, and suggests the feasibilities of ensemble classification methods.
- 133
- 130
- Learning a Proposal Classifier for Multiple Object Tracking128
A novel proposal-based learnable framework, which models MOT as a proposal generation, proposal scoring and trajectory inference paradigm on an affinity graph, and can solve the MOT problem in a data-driven way.
- 127
- Transferred Dimensionality Reduction126
This paper proposes an algorithm named Transferred Discriminative Analysis, which uses clustering to generate class labels for the target unlabeled data, and use dimensionality reduction for them joint with prior labeled data to do subspace selection.
- Adaptive feature extraction for EEG signal classification122
This paper addresses the problem from the aspect of updating feature extractors and proposes an adaptive feature extractor, namely adaptive common spatial patterns (ACSP), which through the weighed update of signal covariances extracts the most discriminative features related to the current brain states.
- Evolutionary hierarchical dirichlet processes for multiple correlated time-varying corpora120
This paper forms the EvoHDP as a series of hierarchical Dirichlet processes~(HDP) by adding time dependencies to the adjacent epochs, and proposes a cascaded Gibbs sampling scheme to infer the model.
- 119
- Seeing the Arrow of Time116
Good video forwards/backwards classification results are demonstrated on a selection of YouTube video clips, and on natively-captured sequences (with no temporally-dependent video compression), and what motions the models have learned that help discriminate forwards from backwards time are examined.
- 116
- 115
- 114
- Competitive EM algorithm for finite mixture models111
A novel competitive EM algorithm for finite mixture models to overcome the two main drawbacks of the EM algorithm: often getting trapped at local maxima and sometimes converging to the boundary of the parameter space is presented.
- Graph based multi-modality learning110
This paper reveals the regular optimization nature, transductive learning nature as well as prior fusion nature of the proposed schemes, respectively, and can be easily extended to unsupervised learning, including clustering and embedding.
- 106
- Neighborhood MinMax projections103
A new algorithm, Neighborhood MinMax Projections (NMMP), is proposed for supervised dimensionality reduction, and focuses only on the pairwise points where the two points are neighbors of each other, which has three significant advantages.
- Boosting Semantic Human Matting With Coarse Annotations102
This paper trains a mask prediction network to estimate the coarse semantic mask using the hybrid data, and proposes a quality unification network to unify the quality of the previous coarse mask outputs to boost end-to-end semantic human matting without trimaps as extra input.
- 101
- 100
- Source-Free Domain Adaptation via Target Prediction Distribution Searching98
This paper proves that in case of sufficient small distribution shift, the domain transfer error could be well bounded, and proposes a novel Target Prediction Distribution Searching (TPDS) paradigm, which achieves new state-of-the-art performance.
- A survey on machine learning from few samples98
This survey extensively review 300+ papers of FSL spanning from the 2000s to 2019 and provides a timely and comprehensive survey for FSL, which categorize FSL approaches into the generative model based and discriminative modelBased kinds in principle, and emphasize particularly on the meta learning based F SL approaches.
- 97
- 96
- Computer Vision for Biomedical Image Applications96
Applying Prior Knowledge in the Segmentation of 3D Complex Anatomic Structures for Medical Image Processing is applied.
- A general kernelization framework for learning algorithms based on kernel PCA95
This paper gives a rigorous justification to reveal that under some mild conditions, the kernelization under this framework is equivalent to traditional kernel method, and shows that these mild conditions are usually satisfied in most of learning algorithms.
- 91
- 87
- Reply networks on a bulletin board system86
It is indicated that the hierarchical and clustering structure of the interest space, together with overlapping interests of IDs not only result in small-world characteristics of reply networks on BBS, but also give rise to preferential attachment, which is a popular explanation for scale-free features.
- 82
- Traffic Flow Forecasting Using a Spatio-temporal Bayesian Network Predictor81
A novel predictor for traffic flow forecasting, namely spatiotemporal Bayesian network predictor, is proposed, which incorporates all the spatial and temporal information available in a transportation network to carry the traffic flow forecast of the current site.
- Multiple rank multi-linear SVM for matrix data classification79
It is revealed that the rank of transformation can be regarded as a tradeoff parameter to balance the capacity of learning and generalization in essence and an effective approach to solve the proposed non-convex optimization problem is proposed.
- Finding image exemplars using fast sparse affinity propagation78
A novel approach to organize image search results obtained from state-of-the-art image search engines by adopting the idea of affinity propagation and design a fast sparse affinity propagation algorithm to find exemplars that best represent the imagesearch results.
- 78
- Extracting the optimal dimensionality for local tensor discriminant analysis77
This paper proposes an algorithm to extract the optimal dimensionality for local tensor discriminant analysis and results on a toy example and real-world data validate the effectiveness of the proposed method.
- Efficient multiclass maximum margin clustering77
Experimental evaluations show that the proposed cutting plane algorithm for multiclass maximum margin clustering converges much faster than existing MMC methods with guaranteed accuracy, and can thus handle much larger datasets efficiently.
- Stable local dimensionality reduction approaches77
A unified framework is proposed that explicitly unfolds the manifold and reformulate local approaches as the semi-definite programs instead of the above-mentioned eigenproblems.
- 76
- 71
- ViTAS: Vision Transformer Architecture Search69
This paper develops a new cyclic weight-sharing mechanism for token embeddings of the ViTs, which enables each channel could more evenly contribute to all candidate architectures, and proposes identity shifting to alleviate the many-to-one issue in superformer.
- Reconstruction and analysis of multi-pose face images based on nonlinear dimensionality reduction68
Methods to establish the process of mapping from low-dimensional embedded space to high-dimensional space for LLE and validate their efficiency with the application of reconstruction of multi-pose face images are proposed and proposed.
- 67
- Connectionist Temporal Classification with Maximum Entropy Regularization67
This work proposes a regularization method based on maximum conditional entropy which penalizes peaky distributions and encourages exploration and introduces an entropy-based pruning method to dramatically reduce the number of CTC feasible paths by ruling out unreasonable alignments.
- Orthogonal vs. uncorrelated least squares discriminant analysis for feature extraction66
It is shown that the orthogonal least squares discriminant analysis is an extension to the null space linear discriminative analysis, and the uncorrelated least squares discrimination analysis is exactly equivalent to the traditional linear discriminantAnalysis.
- 65
- Semi-Supervised Classification with Universum64
This paper proposes a graph based method to make use of the Universum data to help depict the prior information for possible classifiers and shows that the proposed method can obtain superior performances over conventional supervised and semi-supervised methods.
- 62
- 61
- Interactive localized content based image retrieval with multiple-instance active learning59
Two general multiple- instance active learning methods are proposed, multiple-instance active learning with a simple margin strategy (S-MIAL) and multiple- instances activeLearning with fisher information (F-MIAl), and apply them to the active learning in localized content based image retrieval (LCBIR).
- Prioritized Architecture Sampling with Monto-Carlo Tree Search58
This paper introduces a sampling strategy based on Monte Carlo tree search (MCTS) with the search space modeled as a Monte Carlo trees (M CT), which captures the dependency among layers and significantly improves search efficiency and performance.
- RL-Duet: Online Music Accompaniment Generation Using Deep Reinforcement Learning58
A deep reinforcement learning algorithm for online accompaniment generation, with potential for real-time interactive human-machine duet improvisation, based on a well-functioning reward model that considers the compatibility of the machine-generated note with both the machine's context and the human-generated context.
- Efficient Maximum Margin Clustering via Cutting Plane Algorithm57
This paper constructs a nested sequence of successively tighter relaxations of the original MMC problem, and each optimization problem in this sequence could be efficiently solved using the constrained concave-convex procedure (CCCP).
- 56
- Multitask Bregman clustering56
In this paper, a general framework for this problem is proposed and a specific approach is suggested, based on the general Bregman divergence, suitable for a large family of assumptions on data distributions and divergences.
- 55
- 55
- A Dexterous Hand-Arm Teleoperation System Based on Hand Pose Estimation and Active Vision53
This work develops a novel vision-based hand-arm teleoperation system that captures the human hands from the best viewpoint and at a suitable distance and ensures the high accuracy of the proposed neural network.
- 53
- marginFace: A novel face recognition method by average neighborhood margin maximization53
An automatic approach for determining the optimal dimensionality of the embedded subspace of the ANMM and the experimental results of applying marginFace to face recognition are presented to show the effectiveness of the method.
- Hierarchical Shape Modeling for Automatic Face Localization52
A novel hierarchical data-driven Markov chain Monte Carlo (HDDMCMC) approach is proposed to achieve the global optimum of face localization and demonstrates that the algorithm produces accurate localization results quickly, bypassing the need for good initialization.
- 51
- Regularized clustering for documents51
This paper proposes a novel method for clustering documents using regularization, called Clustering with Local and Global Regularization (CLGR), and shows that the cluster memberships of the documents can be achieved by eigenvalue decomposition of a sparse symmetric matrix.
- 51
- 50
- Selecting informative universum sample for semi-supervised learning49
This research found that not all the Universum samples are helpful, and a method to pick the informative ones, i.e., in-between universum samples is proposed, which outperforms the former works.
- Embedding new data points for manifold learning via coordinate propagation49
This work first formulate this task as a quadratic programming, and then develops an iterative algorithm for coordinate propagation, to propagate the known coordinates to each of the new data points.
- 49
- 48
- Learning Kernels with Radiuses of Minimum Enclosing Balls48
This paper uses the ratio between the margin and the radius of the minimum enclosing ball to measure the goodness of a kernel, and presents a new minimization formulation for kernel learning that is invariant to scalings of learned kernels and to the types of norm constraints on combination coefficients.
- 48
- Intelligent Science and Intelligent Data Engineering47
This paper gives a continuous objective function based on gradient fields and discusses how to discretize it by using tangent space and shows the geometric meaning of the approach, which is requiring the gradient field as orthogonal as possible to the tangent spaces.
- Efficient Nonnegative Matrix Factorization via projected Newton method47
This paper proposes a fast NMF algorithm via Projected Newton Method (PNM) to efficiently solve a nonnegative least squares problem, which achieves a quadratic convergence rate under appropriate assumptions.
- 45
- Mean version space: a new active learning method for content-based image retrieval45
This paper proposes a novel active learning method named mean version space, aiming to select the optimal image in each round of relevance feedback, and proposes a new criterion which will lead to the fastest shrinkage of the version space in all cases.
- CATRO: Channel Pruning via Class-Aware Trace Ratio Optimization44
A novel channel pruning method via class-aware set function maximization (CATRO) to reduce the computational burden and accelerate the model inference, which achieves higher accuracy with similar computation cost or lower computation cost with similar accuracy than other state-of-the-art channel pruning algorithms.
- 44
- Front-view vehicle detection by Markov chain Monte Carlo method44
Experimental results show that the new vehicle detection approach based on Markov chain Monte Carlo has a high detection rate on vehicles and can perform successful segmentation, and reduce the influence caused by vehicle occlusion.
- The random electrode selection ensemble for EEG signal classification44
A simple yet effective ensemble approach for electroencephalogram (EEG) signal classification named the random electrode selection ensemble (RESE) is developed, which aims to surmount the instability demerit of the Fisher discriminant feature extraction for BCI applications.
- Sparse Coding with Gated Learned ISTA43
This paper introduces overshoot gates for compensating insufficient step size in LISTA and specific design of the gates is inspired by convergence analyses of the mechanism and hence its effectiveness can be formally guaranteed.
- Hierarchical Interactions Model for Predicting Mild Cognitive Impairment (MCI) to Alzheimer's Disease (AD) Conversion43
This paper proposes to fit the prediction models using pairwise biosignature interactions, thus capturing higher-order relationship among biosignatures, and discovers several significant interactions predictive of MCI-to-AD conversion.
- Semi-blind source extraction for fetal electrocardiogram extraction by combining non-Gaussianity and time-correlation43
An objective function for extraction of temporally correlated sources based on the non-Gaussianity and the autocorrelations of source signals and it contains the well-known mean squared error objective function presented by Barros and Cichocki.
- BCNet: Searching for Network Width with Bilaterally Coupled Network42
This paper introduces a new supernet called Bilaterally Coupled Network (BCNet), which is fairly trained and responsible for the same amount of network widths, thus each network width can be evaluated more accurately and turns out to further boost the performance of NAS models by refining their networkwidths.
- Explaining Algorithmic Fairness Through Fairness-Aware Causal Path Decomposition42
This paper considers the causal relationships among feature variables and proposes a novel framework to decompose the disparity into the sum of contributions from fairness-aware causal paths, which are paths linking the sensitive attribute and the final predictions, on the graph.
- 42
- Pseudo Relevance Feedback Based on Iterative Probabilistic One-Class SVMs in Web Image Retrieval42
This paper proposes a novel pseudo relevance feedback method named iterative probabilistic one-class SVMs to re-rank the retrieved images by assuming that most top-ranked images are relevant to the query.
- 41
- Face recognition using spectral features41
A new unsupervised nonlinear feature extraction method called spectral feature analysis (SFA) is proposed, which can extract discriminatory information from the data, and it is shown that linear discriminant analysis can be subsumed under the SFA framework.
- 40
- 40
- 39
- Single image dehazing in inhomogeneous atmosphere38
An effective and robust algorithm based on dark channel prior and the authors' optical model in inhomogeneous atmosphere to remove the haze effect from a single input image and it is demonstrated that this method yields similar or even better results than the state-of-the-art techniques while performing fast.
- Manifold-Ranking-Based Keyword Propagation for Image Retrieval38
A novel keyword propagation method for image retrieval based on a recently developed manifold-ranking algorithm that can be performed online during relevance feedback without extra offline training and can be self-improved constantly by means of keyword model update.
- Image Deblurring with Coupled Dictionary Learning37
This paper presents a learning-based framework where the knowledge hidden in huge amounts of available data is explored and exploited for image deblurring, developed under the conceptual framework of coupled dictionary learning.
- 37
- 37
- 36
- What You See is What You Get: Visual Pronoun Coreference Resolution in Dialogues36
This paper formally defines the task of visual-aware pronoun coreference resolution (PCR) and introduces VisPro, a large-scale dialogue PCR dataset, and proposes a novel visual- aware PCR model, VisCoref, for this task and conducts comprehensive experiments and case studies on the dataset.
- Accurate Forgetting for Heterogeneous Federated Continual Learning35
It is found that forgetting biased information is beneficial in this study and a new concept accurate forgetting (AF) is proposed and a novel generative-replay method is developed which selectively utilizes previous knowledge in federated networks.
- Collaboration Equilibrium in Federated Learning35
This paper proposes the concept of benefit graph which describes how each client can benefit from collaborating with other clients and advances a Pareto optimization approach to identify the optimal collaborators and theoretically proves that it can reach a CE from the benefit graph through an iterative graph operation.
- Solving one-class problem with outlier examples by SVM35
In this paper, a new algorithm "SVM-SVDD" is proposed, in which both Support Vector Machine (SVM) and SVDD are used to solve data description problem with negative examples, and the experimental results illustrate that SVM-sVDD outperforms SVDd-neg on both training time and accuracy.
- Active learning with adaptive regularization34
This paper proposes a novel active learning support vector machine algorithm with adaptive model selection that significantly improves the active learning efficiency with comparatively inexpensive computational cost.
- Kernel Trick Embedded Gaussian Mixture Model34
This paper presents a kernel trick embedded Gaussian Mixture Model (GMM), called kernel GMM, which can solve problems in probabilistic framework and tackle nonlinear problems better than the traditional GMM.
- Reborn Filters: Pruning Convolutional Neural Networks with Limited Data33
This paper proposes to use all original filters to directly develop new compact filters, named reborn filters, so that all useful structure priors in the original filters can be well preserved into the pruned networks, alleviating the performance drop accordingly.
- Semi-supervised discriminative classification with application to tumorous tissues segmentation of MR brain images33
A discriminative classification algorithm for semi-automated segmentation of brain tumorous tissues using non-parametric Bayesian Gaussian random field in the semi-supervised mode is implemented.
- 33
- 32
- 32
- Multi-view local learning32
This paper designs a multi-view local model for each example, proposes a Multi-View Local Learning Regularization (MVLL-Reg) matrix, and designs a linear and kernel version of this matrix.
- Semi-supervised sub-manifold discriminant analysis31
A semi-supervised sub-manifold discriminant analysis algorithm that can give competitive results in comparison to the state-of-the-art algorithms is presented.
- 30
- 30
- Robust non-negative matrix factorization via joint sparse and graph regularization for transfer learning30
This paper proposes a robust non-negative matrix factorization via sparse regularization model (RSNMF) to handle source domain data and then learns a meaningful matrix, which contains much common information between source domain and target domain data.
- Instance-level semisupervised multiple instance learning30
This paper proposes a new graph-based semi-supervised learning approach for multiple instance learning by defining an instance-level graph on the data, and empirically shows that this method outperforms state-of-the-art MIL algorithms on several real-world data sets.
- Bayesian Network Methods for Traffic Flow Forecasting with Incomplete Data30
Experiments show that the proposed Bayesian network methods are applicable and effective for traffic flow forecasting with incomplete data.
- 30
- Leveraging mixed and incomplete outcomes via reduced-rank modeling29
A mixed-outcome reduced-rank regression is developed, which effectively enables information sharing among different prediction tasks and is demonstrated by simulation studies and an application on predicting health-related outcomes in longitudinal studies of aging.
- Semi-supervised ranking aggregation29
A semi-supervised ranking aggregation method, in which the order of several item pairs are labeled as side information, to learn a ranking function based on the ordering agreement of different rankers.
- 29
- 29
- 29
- Gabor face recognition by multi-channel classifier fusion of supervised kernel manifold learning28
A face recognition framework under the multi-channel fusion strategy for Gabor wavelet endows the algorithm in a similar way as the human visual system, to represent face features, and the manifold learning is applied to model the nonlinear labeled intrinsic structure.
- Efficient Euclidean projections via Piecewise Root Finding and its application in gradient projection28
This paper proposes a unified method called Piecewise Root Finding (PRF) to efficiently calculate Euclidean projections onto three typical constraint sets: @?"1-ball, Elastic Net (EN) and the Intersection of a Hyperplane and a Halfspace (IHH).
- 28
- Semi-definite Manifold Alignment28
This method provides a more flexible way to acquire the prior knowledge for alignment, thus is able to handle situations where corresponding pairs are hard or impossible to identify.
- 27
- 27
- An optimal kernel feature extractor and its application to EEG signal classification27
An optimal nonlinear feature extractor for extracting energy features under two different kinds of patterns carries out the simultaneous diagonalization of two signal covariance matrices in a high-dimensional kernel transformed space, and thus promises to find features which are more discriminant, especially when the original data have nonlinear structures.
- Weighted competition scale-free network27
A model of weighted scale-free networks incorporating a fit-gets-richer scheme which means the connectivity of the node depends on both the degree and fitness of the nodes, which indicates that asymptotically the scaling behaviors of the total weight distribution and the connectivity distribution are identical.
- Single-Cell Transcriptomics Reveals Peripheral Immune Responses in Anti-Synthetase Syndrome-Associated Interstitial Lung Disease26
The scRNA-seq data analyses reveal that ASS-ILD is characterized by upregulated interferon responses, altered CD8 T cell homeostasis, and involvement of differentiation signaling pathways of CD4 T cells.
- Rethinking Audio-Visual Synchronization for Active Speaker Detection26
This work proposes a cross-modal contrastive learning strategy and applies positional encoding in attention modules for supervised ASD models to leverage the synchronization cue and suggests that this model can successfully detect unsyn-chronized speaking as not speaking, addressing the limitation of current models.
- Separation of Music Signals by Harmonic Structure Modeling26
A new music signal separation method is proposed, which is based on harmonic structure modeling, and results show that the algorithm can separate signals and obtain not only a very high Signal-to-Noise Ratio (SNR) but also a rather good subjective audio quality.
- Tunable Nearest Neighbor Classifier26
A tunable nearest neighbor (TNN) classifier is proposed to handle the discrimination problems, and consistently achieves better performance than NFL and conventional nearest neighbor methods, especially for the tasks with small training sets.
- 26
- GreedyNASv2: Greedier Search with a Greedy Path Filter25
This paper leverages an explicit path filter to capture the characteristics of paths and directly filter those weak ones, so that the search can be thus implemented on the shrunk space more greedily and efficiently.
- 25
- 25
- Learning a subspace for clustering via pattern shrinking25
A novel subspace learning method is proposed that not only characterizes the linear and nonlinear structures of data, but also reflects the requirements of following clustering and can derive a subspace that is more suitable for high dimensional data clustering.
- 25
- Semi-supervised classification using local and global regularization24
This paper proposes a semi-supervised learning algorithm based on local and global regularization that constructs a regularized classifier for each data point using its neighborhood, and adopts a Laplacian regularizer to smooth the data labels predicted by those local classifiers.
- Semi-supervised metric learning by maximizing constraint margin24
This paper considers the problem of learning a proper distance metric under the guidance of some weak supervisory information in the form of pairwise constraints which specify whether a pair of data points are in the same class (must link constraints) or in the different classes (cannot link constraints).
- Classification of gene-expression data: The manifold-based metric learning way24
A novel metric learning algorithm is proposed by utilizing both local manifold structural information and local discriminant information that can achieve promising results, and gain significant performance improvement over traditional IBL algorithms.
- 23
- A general framework for transfer sparse subspace learning23
An effective sparse regularization is introduced to the proposed transfer subspace learning framework, which can reduce time and space cost obviously, and more importantly,Which can avoid or at least reduce over-fitting problem.
- Subspace ensembles for classification23
This paper probes into the microcosmic meaning of ensemble diversity, and proposes to use region partitioning and region weighting to implement effective subspace ensembles for classification by manipulating different feature subspaces.
- 22
- Cuts3vm22
It is proved theoretically that the CutS3VM algorithm takes time O(sn) to converge with guaranteed accuracy, where n is the total number of samples in the dataset and s is the average number of non-zero features, i.e the sparsity.
- Robust self-tuning semi-supervised learning22
This paper proposes a robust self-tuning graph-based SSL method, which can determine the similarities between pairwise data points automatically and is not sensitive to outliers.
- Intention-Related Natural Language Grounding via Object Affordance Detection and Intention Semantic Extraction21
This work presents an attention-based multi-visual features fusion network to detect object affordances from RGB images, and introduces an intention semantic extraction module to extract intention semantics from intention-related natural language queries.
- Knowledge transfer on hybrid graph21
A transfer learning framework based on similarity matrix approximation to tackle machine learning problems, which can make maximum use of all available similarity information across domains and build a hybrid graph based on all available data.
- Gaussian moments for noisy complexity pursuit21
This paper derives a simple algorithm combining Gaussian moments and complexity pursuit for noisy ICA when Gaussian noise is present and the independent components are time dependent.
- Active curve axis Gaussian mixture models21
Experiments show that the proposed nonlinear mixture models can approximate distributions of data clouds with curve manifolds in a more concise and compact way than GMM does.
- Balancing Similarity and Complementarity for Federated Learning20
This work introduces a novel framework, \texttt{FedSaC}, which balances similarity and complementarity in FL cooperation, and aims to approximate an optimal cooperation network for each client by optimizing a weighted sum of model similarity and feature complementarity.
- 20
- On the Sample Complexity of Random Fourier Features for Online Learning20
This empirical study verifies that online kernel learning with a limited number of random Fourier features can achieve similar generalization performance as online learning using full kernel matrix.
- 20
- 20
- Learning to rank tags20
A novel semi-supervised learning framework to rank image tags is presented, which learns a ranking projection with theoretic guarantee from visual words distribution to the relevant tags distribution, and then uses it for ranking new image tags.
- Multi-view EM Algorithm for Finite Mixture Models20
The proposed Multi-View Expectation and Maximization algorithm for finite mixture models is proposed by us to handle real-world learning problems which have natural feature splits and has these impressing advantages comparing with other algorithms in Co-training setting.
- Exploring the structure of supervised data by Discriminant Isometric Mapping20
This paper introduced a novel algorithm named supervised isometric mapping (SIsomap) which was based on a combination of two well-known methods: isomap and fuzzy linear discriminant analysis (LDA).
- 19
- Recent Advances in Large Margin Learning19
A survey of recent advances in large margin training and its theoretical foundations for (nonlinear) deep neural networks (DNNs) that are probably the most prominent machine learning models for large-scale data in the community over the past decade.
- On Connections Between Regularizations for Improving DNN Robustness19
This paper analyzes regularization terms proposed recently for improving the adversarial robustness of deep neural networks (DNNs), from a theoretical point of view, and investigates possible connections between several effective methods, including input-gradient regularization, Jacobian regularizations, curvature regularization and a cross-Lipschitz functional.
- What if the Irresponsible Teachers Are Dominating?19
A method for picking out the good teachers with promising experimental results is proposed, which works even when the irresponsible teachers are dominating in numbers.
- What if the irresponsible teachers are dominating? a method of training on samples and clustering on teachers19
A method for picking out the good teachers with promising experimental results is proposed, which works even when the irresponsible teachers are dominating in numbers.
- Learning On-Line Classification via Decorrelated LMS Algorithm: Application to Brain-Computer Interfaces19
This paper addresses the challenge of neurophysiological signal classification arising from brain-computer interface (BCI) applications and proposes an on-line classifier designed via the decorrelated least mean square (LMS) algorithm, based on a Bayesian classifier with Gaussian mixture models.
- VD-PCR: Improving visual dialog with pronoun coreference resolution18
This paper proposes VD-PCR, a novel framework to improve Visual Dialog understanding with Pronoun Coreference Resolution in both implicit and explicit ways, and achieves state-of-the-art experimental results on the VisDial dataset.
- 18
- FMA-ETA: Estimating Travel Time Entirely Based on FFN with Attention18
Results show FMA-ETA is competitive with state-of-the-art methods in terms of the prediction accuracy with significantly better inference speed and the novel Multi-factor Attention mechanism is proposed to deal with different category features and aggregate the information purposefully.
- 18
- 18
- 18
- A Random Walks Method for Text Classification18
A random walks method for text classification is proposed, in which the classification problem is formulated as solving the absorption probabilities of Markov random walks on a weighted graph, and the Laplacian operator for asymmetric graphs is derived and utilized for asymmetrical transition matrix.
- Extended Nearest Feature Line Classifier18
An extended nearest feature line (ENFL) classifier is proposed to handle the discrimination problems and consistently achieves better performance than NFL and conventional nearest neighbor methods.
- Road Network Metric Learning for Estimated Time of Arrival17
The Road Network Metric Learning framework for ETA (RNML-ETA) is proposed, which consists of a main regression task to predict the travel time, and an auxiliary metric learning task to improve the quality of link embedding vectors.
- 17
- Contour graph based human tracking and action sequence recognition17
A new framework for human contour tracking and action sequence recognition is introduced, given a gallery of labeled human contours, where each contour is defined as a ''word'' and encoded into a contour dictionary, which is used to translate the video.
- Preshooting Electroencephalographic Activity of Professional Shooters in a Competitive State16
The results showed a significant linear correlation between shooting accuracy and EEG power of anterior frontal, central, temporal, and occipital regions in beta and theta bands, which reveals the changes of cortical activity underlying competition shooting and provides further understanding of the neural mechanisms of the shooting process.
- 16
- Incremental multiple instance outlier detection16
It is proved theoretically that Inc I- MLOF achieves the equal result to that of I-MLOF, an incremental MI outlier detection algorithm that achieves good performance on several synthetic and real data sets.
- 16
- 15
- Towards Model-Agnostic Post-Hoc Adjustment for Balancing Ranking Fairness and Algorithm Utility15
Comprehensive experiments on a suite of benchmark data sets and two real-world patient electronic health record repositories show that the proposed model agnostic post-processing framework can achieve a great balance between the algorithm utility and ranking fairness.
- Adversarial Margin Maximization Networks15
This paper proposes adversarial margin maximization (AMM), a learning-based regularization which exploits an adversarial perturbation as a proxy and encourages a large margin in the input space, just like the support vector machines.
- Knowledge Distillation from Few Samples15
A novel and simple method for knowledge distillation from few samples by adding a 1x1 conv-layer at the end of each block in the student-net, and align the block-level outputs between "teacher" and "student" by estimating the parameters of the added layer with limited samples.
- Relaxed sparse eigenvalue conditions for sparse estimation via non-convex regularized regression15
The conditions of sparse estimations for the sharp concave regularizers which are a general family of non-convex regularizers including many existing regularizers are studied to prove that the desired AGAS solutions can be obtained by coordinate descent (CD) based methods.
- Learning similarity metric with SVM15
A novel approach to simultaneously learn a Mahalanobis similarity metric and an SVM classifier is presented, which generalizes the traditional large margin principle used in standard SVM, that is, it maximize the margin-radius-ratio.
- 15
- 15
- 15
- Automatically Predicting Material Properties with Microscopic Images: Polymer Miscibility as an Example14
This work achieves automatic miscibility recognition utilizing a convolutional neural network and transfer learning methods, and the model obtains up to 94% accuracy.
- Bipartite Ranking Fairness Through a Model Agnostic Ordering Adjustment14
This article proposes a model agnostic post-processing framework xOrder for achieving fairness in bipartite ranking and maintaining the algorithm classification performance, and optimize a weighted sum of the utility as identifying an optimal warping path across different protected groups and solve it through a dynamic programming process.
- Searching for Network Width With Bilaterally Coupled Network14
This article introduces a new supernet called Bilaterally Coupled Network (BCNet) and proposes a new open-source width search benchmark on macro structures named Channel-Bench-Macro for the better comparisons of the width search algorithms with MobileNet- and ResNet-like architectures.
- Attention Based Visual Analysis for Fast Grasp Planning With a Multi-Fingered Robotic Hand14
Experimental results show that the proposed framework can not only speed up grasp planning with more stable configurations, but also handle unknown objects.
- 14
- Image deblurring with matrix regression and gradient evolution14
This paper presents a supervised learning algorithm for image deblurring that learns a matrix mapping to map each of its image patches directly to be a new one with more sharp details to be deblurred in way of gradient evolution.
- Boosting with pairwise constraints14
A novel framework to solve the problem of extending AdaBoost to semi-supervised scenarios based on the gradient descent view of boosting is proposed, which is almost as simple and flexible as AdaBoost, and can be readily applied in the presence of pairwise constraints.
- 14
- Finite Mixture Models with Negative Components14
By cancelling the nonnegative constraint to mixture coefficients and introducing a new concept of negative components, this work extends the traditional mixture models and enhance their performance without increasing the complexity obviously.
- Boosting GMM and Its Two Applications14
This paper generalizes the framework of boosting to a specific density model – Gaussian Mixture Model (GMM) and proposes the boosting GMM algorithm, which will illustrate the applications of the algorithm to cluster ensemble and short-term traffic flow forecasting problems.
- 14
- When Counterpoint Meets Chinese Folk Melodies13
This paper proposes a reinforcement learning-based system, named FolkDuet, towards the online countermelody generation for Chinese folk melodies, with no existing data of Chinese folk duets, and employs two reward models based on out-of-domain data.
- On-line evolutionary exponential family mixture13
All conventional clustering algorithms based on EFMs can be extended to evolutionary setting under the two frameworks, which are based on the general exponential family mixture (EFM) model.
- 13
- Fast Multilevel Transduction on Graphs13
This paper proposes a multilevel scheme for speeding up the traditional graph based semi-supervised learning methods, which has explicit physical meanings with some graph intuitions and provides a theoretical guarantee of the performance of the method.
- Physics Reasoner: Knowledge-Augmented Reasoning for Solving Physics Problems with Large Language Models12
Physicially, Physics Reasoner mitigates the issues of insufficient knowledge and incorrect application, achieving state-of-the-art performance on SciBench with an average accuracy improvement of 5.8%.
- 12
- 12
- Constructing Geographic and Long-term Temporal Graph for Traffic Forecasting12
This work proposes Geographic and Long-term Temporal Graph Convolutional Recurrent Neural Network (GLT-GCRNN), a novel framework for traffic forecasting that learns the rich interactions between roads sharing similar geographic or longterm temporal patterns.
- On the Generalization Ability of Online Gradient Descent Algorithm Under the Quadratic Growth Condition12
This paper studies the generalization ability of the classic online gradient descent algorithm under the quadratic growth condition (QGC) and indicates that the strong convexity is not necessary to achieve the sharp convergence rate in online learning.
- Large-scale eigenvector approximation via Hilbert Space Embedding Nyström12
A novel criterion named Hilbert Space Embedding Nystrom criterion is proposed which directly minimizes the eigenvector approximation error and is shown to provide a general framework to approximate eigenvectors within linear time and space complexity.
- A multilevel approach for learning from labeled and unlabeled data on graphs12
This paper proposes a multilevel scheme for speeding up the traditional graph based semi-supervised learning methods that has explicit physical meanings with some graph intuitions and has guaranteed performance since it is closely related to the algebraic multigrid methods.
- 12
- 12
- Probabilistic tangent subspace: a unified view12
This paper extends TD to more general pattern classification tasks, and proposes three probabilistic subspace models to encode the variations: the linear sub space, nonlinear subspace, and manifold sub space models.
- 11
- 11
- 11
- Online kernel learning with nearly constant support vectors11
This work studied online regularized kernel learning using the Nystrom method, with a focus on the sample complexity, and shows that, when the loss function is smooth and strongly convex, only O randomly sampled support vectors are needed to guarantee an O ( log T / T ) convergence rate, which is almost optimal except for the log T factor.
- Semi-supervised classification on evolutionary data11
This paper discusses how to carry out temporal smoothness assumption using temporal regularizers defined in a structural way with respect to the Hilbert space, and derives the online algorithm that efficiently finds the closed-form solution to the classification functions.
- Fast nonlinear autocorrelation algorithm for source separation11
Using a statistical property-nonlinear autocorrelation and maximizing the nonlinear autOCorrelation of source signals, a fast fixed-point algorithm for BSS is proposed and it is shown that its convergence speed is at least quadratic.
- 11
- Nonlinear innovation to blind source separation11
Simulations verify the efficient implementation of the proposed blind source separation method, especially its robustness to the outliers.
- Semi-Supervised Mean Fields11
A novel semi-supervised learning approach based on statistical physics that treats each data point as an Ising spin and the interaction between pairwise spins is captured by the similarity between the pairwise points.
- Spline Embedding for Nonlinear Dimensionality Reduction11
A new algorithm for nonlinear dimensionality reduction (NLDR) where smoothing splines are used to map the locally-coordinatized data points into a single global coordinate system of lower dimensionality by minimizing the low-dimensional coordinate reconstruction error.
- Socratic Questioning: Learn to Self-guide Multimodal Reasoning in the Wild10
This work devise an innovative multi-round training and reasoning framework suitable for lightweight Multimodal Large Language Models (MLLMs), named Socratic Questioning (SQ), which heuristically guides MLLMs to focus on visual clues relevant to the target problem, reducing hallucinations and enhancing the model's ability to describe fine-grained image details.
- Subjective Topic meets LLMs: Unleashing Comprehensive, Reflective and Creative Thinking through the Negation of Negation10
A framework grounded in the principle of the Negation of Negation (NeoN) is introduced to unleash the potential comprehensive, reflective, and creative thinking abilities of LLMs.
- 10
- 10
- Efficient Active Learning with Boosting10
An efficient active learning algorithm is derived, based on a novel query mechanism called query by incremental committee, which does not only save considerable computational cost, but also outperforms conventional active learning methods based on boosting.
- Efficient multi-class unlabeled constrained semi-supervised SVM10
A general kernel version semi-supervised dual coordinate descent algorithm is put forward to efficiently solve MCUCSVM and makes it more applicable to problems with large number of classes and large scale labeled data.
- 10
- 10
- 10
- 9
- A faster cutting plane algorithm with accelerated line search for linear SVM9
A novel linear-time line search solver which spends only linear time is proposed while the existing strategy spends O(mlogm) time and an optimized explicit piecewise linear function finding algorithm for multiclass linear SVM is derived.
- A novel unsupervised approach to discovering regions of interest in traffic images9
The study shows that the distribution of such data can be captured by a simplex in a linear subspace, and each data point can be represented by a linear reconstruction over the set of vertices of the simplex.
- Learning high-dimensional correspondence via manifold learning and local approximation9
A new framework for high-dimensional correspondence learning is summarized, and within this framework, a new approach, Local Approximation Maximum Variance Unfolding, could achieve higher accuracy than other machine learning–based methods.
- On discriminative semi-supervised classification9
This paper argues that rather than penalizing the label smoothness, one can directly punish the discriminality of the classification function to achieve a more powerful predictor, and derives two specific algorithms: Semi-Supervised Discriminative Regularization (SSDR) and Semi-parametric Discriming Semi-supervised Classification (SDSC).
- Combining stroke-based and selection-based relevance feedback for content-based image retrieval9
This work proposes a flexible interaction mechanism for CBIR by enabling relevance feedback inside images through drawling strokes, and implements the system Flexible Image Search Tool (FIST) based on the ideas above.
- Session Identification Based on Time Interval in Web Log Mining9
This paper calculates the time intervals of page views, and analyzes the intervals to obtain a certain threshold, which is then used to break the web logs into sessions, and some IPs with special features of frequency distributions can be deemed as single users.
- Multiple random walk and its application in content-based image retrieval9
MRW can be seen as a transductive learning method, which has been demonstrated to outperform inductive ones by previous research work, and systematic experiments on a general-purpose image database consisting of 5,000 Corel images demonstrate the superiority of MRW over state-of-the-art techniques.
- Spectral Clustering for Time Series9
A general framework for time series clustering based on spectral decomposition of the affinity matrix and a gradient based method for self-tuning the variance of the Gaussian function is presented.
- CALM: Consensus-Aware Localized Merging for Multi-Task Learning8
A Consensus-Aware Localized Merging (CALM) method which incorporates localized information aligned with global task consensus, ensuring its effectiveness post-merging and significantly outperforming existing methods and achieving performance close to traditional MTL.
- 8
- 8
- 8
- 8
- Learning Discriminative Piecewise Linear Models with Boundary Points8
A new discriminative piecewise linear model for classification, in which the number of linear surface pieces is automatically determined by the Bayesian inference according to data, is introduced.
- 8
- Generalized additive Bayesian network classifiers8
This paper first generates a series of very simple BNs, put them in the framework of GAM, then adopt a gradient-based algorithm to learn the combining parameters, and thus construct a more powerful classifier.
- 8
- Visual object recognition using probabilistic kernel subspace similarity8
Experiments on synthetic data and real visual object recognition tasks show that probabilistic kernel subspace similarity-based face matching can achieve promising performance, and outperform many other current popular object recognition algorithms.
- On the On-line Learning Algorithms for EEG Signal Classification in Brain Computer Interfaces8
This paper proposes to use the idea of gradient decorrelation to develop the existent basic Least Mean Square (LMS) algorithm for the on-line learning of Bayesian classifiers employed in brain computer interfaces.
- Clustering in Knowledge Embedded Space8
A new algorithm that learns a new distance metric by knowledge embedding and makes clustering become easier under this distance metric and can construct a quality neighborhood graph from a complex and noisy data set and well solve clustering problems.
- Learning without Isolation: Pathway Protection for Continual Learning7
A novel CL framework, learning without isolation (LwI), where model fusion is formulated as graph matching and the pathways occupied by the old tasks are protected without being isolated, is proposed, where the sparsity of activation channels in a deep network can adaptively allocate available pathways for a new task.
- 7
- 7
- Semismooth Newton Algorithm for Efficient Projections onto $\ell_{1, \infty}$-norm Ball7
An efficient algorithm for Euclidean projection onto (cid:96) 1 , ∞ -norm ball is proposed and is promising for the optimization of learning problems with the (cid:96) 1 , ∞ -norm ball constraint.
- 7
- 7
- Localized Content-Based Image Retrieval Using Semi-Supervised Multiple Instance Learning7
A Semi-Supervised Multiple-Instance Learning algorithm is proposed, and applied to Localized Content-Based Image Retrieval (LCBIR), where the goal is to rank all the images in the database, according to the object that users want to retrieve.
- Semi-supervised additive logistic regression: A gradient descent solution7
A semi-supervised regularized method for additive logistic regression where the graph regularization term of the combined functions is added to the original cost functional used in AdaBoost to constrain the learned function to be smooth on a graph.
- 7
- 7
- 7
- Contour Matching Based on Belief Propagation7
This paper uses graphical model based probabilistic inference methods to solve the problem of contour matching, which is a fundamental problem in computer vision and comparable experimental results show that better correspondences can be achieved.
- 7
- On the formation of degree and cluster-degree correlations in scale-free networks7
This paper presents the mean-field rate equation for clustering and obtains analytically the degree-dependence of the cluster-degree, which is size dependent but the tail is kept invariant for different degree exponents in the GBA model.
- 7
- Automatic Peak Number Detection in Image Symmetry Analysis7
This paper designs a new feature to depict image symmetry property which can be automatically extracted for repeated pattern retrieval and demonstrates the effectiveness of the optimization approach and the superiority of symmetry feature over wavelet feature in discriminating similar repeated patterns.
- 7
- Short-Term Traffic Flow Forecasting Using Expanded Bayesian Network for Incomplete Data7
Experiments show that the expanded Bayesian network method is appropriate and effective for short-term traffic flow forecasting with incomplete data.
- 7
- DreamArrangement: Learning Language-Conditioned Robotic Rearrangement of Objects via Denoising Diffusion and VLM Planner6
This work presents DreamArrangement, a novel language-conditioned object rearrangement scheme, consisting of two primary processes: employing a transformer-based multimodal denoising diffusion model to envisage the desired arrangement of objects, and leveraging a vision–language foundational model to derive actionable policies from text, alongside initial and target visual information.
- A Theoretical View of Linear Backpropagation and its Convergence6
It is demonstrated that, somewhat surprisingly, LinBP can lead to faster convergence in neural-network-involved learning tasks, including adversarial attack and model training, in the same hyper-parameter settings, compared to BP.
- Weak Augmentation Guided Relational Self-Supervised Learning6
This paper introduces a novel SSL paradigm, which is term as relational self-supervised learning (ReSSL) framework that learns representations by modeling the relationship between different instances by employing sharpened distribution of pairwise similarities among different instances as relation metric.
- Joint Spatial-Temporal and Appearance Modeling with Transformer for Multiple Object Tracking6
A novel solution named TransSTAM, which leverages Transformer to effectively model both the appearance features of each object and the spatial-temporal relationships among objects, and achieves a clear performance improvement in both IDF1 and HOTA with respect to previous state-of-the-art approaches.
- Editorial: Cross-Modal Learning: Adaptivity, Prediction and Interaction6
This paper presents a meta-modelling system that automates the very labor-intensive and therefore time-heavy and expensive and expensive process of manually cataloging and sorting the molecules in the brain.
- 6
- Self-reinforcing Unsupervised Matching6
This work proposes a dynamic programming algorithm, dynamic position warping (DPW), to reveal the underlying element correspondence relationship between two matrix-form data in an order-preserving fashion, and devise a local feature adapter (LoFA) to allow for cross-modality similarity measurement.
- xOrder: A Model Agnostic Post-Processing Framework for Achieving Ranking Fairness While Maintaining Algorithm Utility.6
This paper proposes a general post-processing framework, xOrder, for achieving fairness in bipartite ranking while maintaining the algorithm classification performance, and formulate this problem as identifying an optimal warping path across different protected groups and solve it through a dynamic programming process.
- DeepDefense: Training Deep Neural Networks with Improved Robustness.6
The core idea is to integrate an adversarial perturbation-based regularizer into the classification objective, such that the obtained models learn to resist potential attacks, directly and precisely.
- 6
- A fast dual projected Newton method for l 1 -regularized least squares6
Experimental results indicate that DPNM is very efficient for solving the l1-regularized least squares problem, by comparing it with state of the art methods.
- 6
- 6
- 6
- Graph Based Multi-Class Semi-Supervised Learning using Gaussian Process6
A multi-class semi-supervised learning algorithm of the graph based method that makes use of the Bayesian framework of Gaussian process to solve the transductive and inductive problems.
- Texture Image Segmentation: An Interactive Framework Based on Adaptive Features and Transductive Learning6
An interactive framework for texture segmentation that the user can define the textures to be segmented by labelling a small part of points belonging to them and further improve the segmentation quality through a few interactive manipulations if necessary.
- Network Game and Boosting6
Comparison results using several datasets of the UCI machine learning repository and synthetic data show that Network Boosting has much resistance to the noisy data than AdaBoost through the cooperation of classifiers in the classifier network.
- 6
- Color Image Segmentation: Kernel Do the Feature Space6
The scheme is that an extension of EM algorithm for GMM is proposed by involving one kernel feature extraction step, which is called K-EM, which not only speeds up kernel step, but also automatically extracts good features for clustering in a nonlinear way.
- 6
- 5
- 5
- 5
- Exophoric Pronoun Resolution in Dialogues with Topic Regularization5
This work proposes to jointly leverage the local context and global topics of dialogues to solve the out-of-text PCR problem and demonstrates the effectiveness of adding topic regularization for resolving exophoric pronouns.
- 5
- Special Article Series on Signal Processing Education via Hands-On and Design Projects [From the Guest Editors]5
As professionals in the signal and information processing field, they build on the tools and theories they learned in their undergraduate studies, adding knowledge and skills over the years to implement new successful teaching methods that advance effective learning and skill development.
- 5
- High-dimensional Inference via Lipschitz Sparsity-Yielding Regularizers5
It is proved that some nonconvex regularizers can be such ”good” regularizers, a family of sparsityyielding penalties with proper Lipschitz subgradients that keep the superiority of non-conveX regularizers in inference.
- A Flexible and Effective Linearization Method for Subspace Learning5
Linear discriminant analysis (LDA), as a supervised algorithm, aims to maximize the inter-class scatter and at the same time minimize the intra- class scatter.
- 5
- 5
- Analysis of classification margin for classification accuracy with applications5
This paper proposes a max-min type bound concerning the minimal misclassification rate, and presents some useful properties of this bound, and devise an algorithm for improving average classification accuracy based on the proposed bound.
- 5
- An evolutionary system for near-regular texture synthesis5
A novel texture synthesis algorithm is proposed that can achieve high-quality synthesized results on a large variety of near-regular textures without any extra overhead for memory and pretreatment, and the speed approaches real-time.
- 5
- 5
- 4
- Study on Allele Specific Expression of Long-Term Residents in High Altitude Areas4
The differential expression of alleles in the high-altitude population and the normal population based on the RNA sequencing data is analyzed and it is realized that the calmodulin homology domain may have a certain correlation with long-term survival at high altitude.
- 4
- 4
- 4
- Multiscale Point Correspondence Using Feature Distribution and Frequency Domain Alignment4
A hybrid scheme is proposed to find the reliable point-correspondences between two images, which combines the distribution of invariant spatial feature description and frequency domain alignment based on two-stage coarse to fine refinement strategy.
- 4
- 4
- 4
- MACBSE: Extracting signals with linear autocorrelations4
The MACBSE approaches are batch fixed-point learning algorithms for extraction of source signals with linear autocorrelations, which are very simple and do not need to choose any learning step sizes.
- Interactive Visual Object Extraction Based on Belief Propagation4
An interactive framework which can allow the user to label multiply foreground objects of interest and some applications in color transfer, image completion and motion detection are given in this paper.
- 4
- Approach to Solve Multiobjective Optimization Problems Based on Immune Principle4
Experimental results indicate that the proposed approach can resolve the problem of multiobjective optimization effectively and has better perfor- mances.
- Mixture Random Effect Model Based Meta-analysis for Medical Data Mining4
A more general statistical model for meta-analysis is proposed to integrate heterogeneous medical researches efficiently and can discover underlying structure and intrinsic heterogeneity of meta datasets, but also can imply reasonable subgroup division.
- Type II Topological Logic $\mathbb{C}^2_\mathcal{T}$ and Approximate Reasoning4
The type II completeness and type II perfectness of knowledge base K is investigated and the structure of type II regular topological logic C 2 T and regular matching function C and C-match neighborhood group is proposed.
- Network Boosting for BCI Applications4
The result on the Data set II of BCI (Brain-computer interface) competition III shows that Network Boosting achieves higher classification accuracy than logistic regression, SVM, Bagging and AdaBoost.
- A New Approach to Human Motion Sequence Recognition with Application to Diving Actions4
This paper proposes an approach to human motion sequence recognition based on 2D spatio-temporal shape analysis, which is used to identify diving actions and shows that high correct recognition ratio can be achieved.
- 4
- Network boosting on different networks4
The analysis based on the synchronizability of network shows that the ensemble learned by scale-free-network-based NB is more correlated than that of NB based on other two topologies.
- 4
- 4
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- Automatic Methods for Screening and Assessing Scoliosis by 2-D Digital Images3
The paper proposes an automatic method to screen and assess scoliosis, especially for the early-stage cases, that only requires 2-D digital images containing human back to reach comparable results with manual results marked on the same images by clinicians.
- Online multiple instance regression3
OnlineMIR can not only predict the label of a new bag, but also update the current regression model with the latest arrived bag, and achieves good performances on both synthetic and real data sets.
- SVM-SVDD: A New Method to Solve Data Description Problem with Negative Examples3
This paper proposes a new approach "SVM-SVDD", in which Support Vector Machine (SVM) helps SVDD to solve data description problem with negative examples efficiently, which obtains its solution by solving two convex optimization problems in two steps.
- 3
- 3
- Type-I Topological Logic $\mathbb{C}^{1}_\mathcal{T}$ and Approximate Reasoning3
The concept of the formulae mass, the knowledge mass and the approximating knowledge closure of the knowledge library by means of topological closure and the fundamental framework of type-I topological logics is obtained.
- Computer vision for biomedical image applications : First International Workshop, CVBIA 2005, Beijing, China, October 21, 2005 : proceedings3
Applying Prior Knowledge in the Segmentation of 3D Complex Anatomic Structures for Medical Image Processing is applied.
- 3
- Computer Vision for Biomedical Image Applications: First International Workshop, CVBIA 2005, Beijing, China, October 21, 2005, Proceedings (Lecture Notes in Computer Science)3
One that the authors will refer to break the boredom in reading is choosing computer vision for biomedical image applications first international workshop cvbia 2005 beijing china october 21 2005 proceedings lecture notes in computer science as the reading material.
- 3
- 3
- Discovery of Relationships between Interests from Bulletin Board System by Dissimilarity Reconstruction3
A new method to analyze people’s interests by simulating the gradually changing and transferring mechanism among interests by putting forward a technique named Dissimilarity Reconstruction (DSR) to discover relationships between the interests.
- 2
- Learned ISTA with Error-Based Thresholding for Adaptive Sparse Coding2
It is shown that the proposed EBT mechanism well disentangles the learnable parameters in the shrinkage functions from the reconstruction errors, endowing the obtained models with improved adaptivity to possible data variations.
- 2
- Structured BFGS Method for Optimal Doubly Stochastic Matrix Approximation2
This paper considers the optimal approximation of a square matrix in the set of doubly stochastic matrices and builds curvature information into the diagonal components of the true Hessian, so that the resulting algorithm outperforms the state-of-the-art solvers and enjoys outstanding scalability.
- Workshop on Model Mining2
This Workshop on Model Mining aims to investigate more diverse and advanced manners in mining knowledge within models, which tends to leverage the pretrained models more wisely, elegantly and systematically.
- 2
- 2
- 2
- Training Deep Belief Network with Sparse Hidden Units2
By introducing sparsity constraint to the Contrastive Divergence algorithm (CD algorithm), a framework to train Restricted Boltzmann Machine (RBM) which is the basic block for Deep Belief Network (DBN).
- Efficient sparse recovery via adaptive non-convex regularizers with oracle property2
An algorithm for efficient sparse recovery based on proximal gradient descent that is compatible with most popular non-convex regularizers, achieves a geometric convergence rate for the recovery error, is selection consistent, and most importantly has the oracle property is developed.
- 2
- Soft Constraint Harmonic Energy Minimization for Transductive Learning and its Two Interpretations2
This paper extends the harmonic energy minimization algorithm and proposes a novel transductive learning algorithm on graph with soft label and soft constraint, which can be interpreted from the view of label propagation and a special random walks on graph, which make the algorithm intuitively reasonable.
- A new approach to discover interlacing data structures in high-dimensional space2
This paper presents a novel approach called Multi-Manifold Partition to identify the interlacing low dimensional patterns and shows that this new approach to exploratory data analysis is effective and may enhance the understanding of the data distribution.
- 2
- 2
- Semi-supervised dimensionality reduction for image retrieval2
Based on the theoretical analysis of the relationship between graph Laplacian and manifold Laplace-Beltrami operator, an online learning algorithm is developed that can incrementally learn the unlabeled data.
- Learning and innovation expand cooperative network topologies2
It is shown that cooperation remains rather stable by applying long-term learning + innovative strategy adoption rules on a variety of random, regular, small-word, scale-free and modular networks in repeated, multi-agent games.
- 2
- Fingerprint Ridge Line Extraction Based on Tracing and Directional Feedback2
Each ridge line is modeled as sequential frames of a continuous curve and then traced by standard CONDENSATION algorithm in the area of computer vision and performs better both in terms of efficiency and robustness.
- A Bayesian Method for High-Frequency Restoration of Low Sample-Rate Speech2
A novel High-frequency (HF) restoration method of low sample-rate speech based on Bayesian inference, which turns the restoration problem into a maximizing a posteriori estimation.
- 2
- On the Scale-Free Intersection Graphs2
It is demonstrated that the degree distribution of terminal vertices are decided only by the attractions decay of the terminal vertice, based on the former results of scale-free intersection graphs.
- Rectified activity-dependent population plasticity implicates cortical adaptation for memory and cognitive functions1
Simulation analysis reveal that the local inhibitory connections might account for the residual potentiation in RAPP, and introducing the RAPP phenomenon in the artificial neural network show promising improvement in small sample size pattern recognition tasks.
- 1
- Automatically Predict Material Properties with Microscopic Image Example Polymer Compatibility1
This work achieves automatic miscibility recognition utilizing convolution neural network and transfer learning method and the model obtains up to 94% accuracy, and puts forward a quantitative criterion for polymer miscibility with this model.
- 1
- 1
- 1
- Efficient Multi-Stage Conjugate Gradient for Trust Region Step1
An efficient Multi-Stage Conjugate Gradient algorithm to compute the trust region step in a multi-stage manner and can generate a solution in any prescribed precision controlled by a tolerance parameter which is the only parameter the authors need.
- 1
- 1
- 1
- 1
- Tree-based compact hashing for approximate nearest neighbor search1
This paper proposes a novel hashing method motivated by maximizing the probability of data with the same hash code being true neighbors, under the constraint of code compactness, which is data-dependent and generates quite compact hash codes.
- Graph-Based Semi-Supervised Learning on Evolutionary Data1
Empirical evaluations show that the GSSLE method outperforms other state-of-the-art methods in terms of stability and is able to deal with dynamic feature space tasks and proves efficient even if the authors do not have many unlabeled samples in the semi-supervised procedure.
- 1
- Learning on Time-Evolving Data1
The recent advance of the research on time-evolving data is surveyed, and some problems worth attention in the future are introduced, such as the mechanism of Time-Evolving Data, generality, classification.
- Exploiting the Limits of Structure Learning via Inherent Symmetry1
This is the first paper to consider the structure learning problem via inherent symmetries of the whole ensemble in Gaussian Markov random fields from i.i.d. samples, and establishes the connection of the learning limit and eigenvalues of concentration matrices, which further leads to a sharper structure learning limit.
- Special issue on the Sino-foreign-interchange workshop on intelligence science and intelligent data engineering1
This special issue is a collection of seven papers on pattern recognition techniques in various application areas, such as document analysis, speech recognition and biometric identification, and has seven papers describing some recent research on computer vision and image processing.
- Non-Asymptotic Analysis of Relational Learning with One Network1
This theoretical paper presents the first probably approximately correct (PAC) bound for maximum likelihood estimation (MLE) and proposes a novel combinational approach to analyze complex dependencies of relational data, which is crucial to this non-asymptotic analysis.
- 1
- Methods for Photomosaic Generation Based on Different Image Similarity and Division Strategies1
A heuristic position searching strategy is proposed to keep the original length-width ratio of each tile in the photomosaic generated unchanged if tiles in the database have different sizes.
- PCA-FA: Applying Supervised Learning to Analyze Gene Expression Data1
A new method is proposed by combining two well-known methods: principal component analysis (PCA) and Fisher analysis (FA) that realizes supervised learning with two types of attributes (nominal attributes and interval- scaled attributes).
- 1
- 1
- 1
- 1
- 1
- 1
- Exploration of Undergraduate Experimental Teaching System and Laboratory Management Mode for Automation Discipline1
The laboratory has built a multi-level, comprehensive and research-based experimental teaching system to educate qualified students in automation discipline with solid foundation as well as broad knowledge and improves the open operation mechanism of laboratories and realizes laboratory resources sharing and maximal efficiency.
- Research on Parameter Automatism Measuring System in Refining Sugar Process1
The main work of this paper is to build a efficient data acquisition application system based on LabVIEW that gives practical selection advices of related sensors and DAQ devices.
- 1
- Distortion-Free Nonlinear Dimensionality Reduction1
This paper shows that the existing Laplacian Eigenmaps method suffers from the distortion problem, and proposes a new distortion-free dimensionality reduction method by adopting a local linear model to encode the local information.
- 1
- 1
- Estimating Fractal Intrinsic Dimension from the Neighborhood1
The proposed method requires lower computation, can reach accurate results and can be flexibly extended, and both the theoretical analysis and experimental results show its validity.
- Using a Random Subspace Predictor to Integrate Spatial and Temporal Information for Traffic Flow Forecasting1
A robust traffic flow predictor is developed integrating the entire spatial and temporal information in a transportation network to cope with this case, and results demonstrate the effectiveness and robustness of the random subspace predictor.
- 1
- 1
- Fault-Tolerant EM Algorithm for GMM in Sensor Networks.1
Performing estimations in a series of subspaces, FEM is able to handle the incomplete data from sensor networks which have some malfunction sensors and efficiently approximates the underlying distribution of sensor data by transferring limited parameters in each communication, which leads to the economization of power consumption.
- Reconfigurable 3D colour information acquisition system based on laser1
The new reconfigurable 3D laser colour information acquisition system has been developed to overcome the problems of the former three-dimensional laser color scanner (3DLCS) and enhances the speed of obtaining 3D information of object greatly and overcome the computational bottleneck.
- 1
- Learning the Supervised NLDR Mapping for Classification1
This work applies neural network to learning an explicit mapping from original space to embedded space and then to classify, and shows that the proposed method has satisfactory performance.
- 1
- A Multi-agent Based Approach to Modelling and Rendering of 3D Tree Bark Textures1
The biological mechanism in tree bark growth is simplified and a MAS model is built to simulate the generation of tree barks to give much fidelity to computer graphics applications.
- 1
- Signature data generation method1
The results of the generated signatures show that the method proposed is useful, and prove that the relations concluded are steady and can be used directly to signature identification.
- Crease's discriminating power in palmprint recognition1
The work in image preprocessings, automatic threshold selection in the key part of system, and crease lines matching are advanced, and an absolute threshold of similarity degree, which ensures all the image pairs with similarity degree above it are of the same palmprint are figured out.
- –
- –
- –
- –
- –
- –
- –
- –
- A Multi-Channel Data Acquisition System for Capacitive Sensing–
A multi-channel data acquisition system for capacitive sensing was designed based on an FDC2214 with external L-C resonators, and good experimental results verified the availability of the designed system.
- –
- –
- –
- –
- –
- –
- –
- –
- Domain Transfer via Multiple Sources Regularization–
A multiple sources regularization framework that extends classification model with regularization by adding a special regularization term, which penalizes the target classifier far from the convex combination of source classifiers simultaneously, and applies to SVM classification model.
- Damping proximal coordinate descent algorithm for non-convex regularization–
The Damping Proximal Coordinate Descent algorithms are proposed that address the optimization issues of a general family of non-convex regularized problems and are guaranteed to be globally convergent.
- –
- A Theory of High-dimensional Sparse Estimation via Non-Convex Regularized Regression.–
This paper provides a general theory to non-convex high-dimensional sparse estimation and can serve as a guideline for selecting regularizers and developing algorithms for non- Convex regularized regression.
- –
- –
- –
- –
- –
- –
- Mining interlacing manifolds in high dimensional spaces–
Experimental results on two synthetic datasets and a practical Optical Character Recognition (OCR) problem show that MMP is effective in extracting interlacing structures and thus offers better interpretation of the nature of the data.
- –
- –
- Compressed Learning with Regular Concept–
The regularity assumption is proposed over a pair of concept and data distribution to greatly generalize former assumptions and defines a robust factor to characterize the margin distribution and shows that such a factor tightly controls the generalization error of a learned classifier.
- –
- –
- Instance- and bag-level manifold regularization for aggregate outputs classification–
A manifold regularization framework is set up to deal with the aggregate outputs classification problem, and four concrete algorithms are proposed, each of which can cope with both binary and multi-class scenarios.
- Application of Siemens Automation to the Bar Strip Mills Modification–
The structure of all continuum rolling mill modification automatic control system in Nangtong ironSteel Co.LTD of BaoSteel has been introduced and has been stably operating.
- –
- –
- Semi-supervised Gaussian random field transduction and induction–
A semi-supervised inductive algorithm adopting a Gaussian random field (GRF) and Gaussian process and a new conditional probability called the extended Bernoulli model (EBM) is proposed, which generalizes the logistic regression to the semi- supervised case.
- Equivalence Knowledge Mass and Approximate Reasoning in $\mathcal{R}$ –Logic $\mathbb{C}_\mathcal{R}$ (I)–
Another matching scheme between the input Ai½ and the knowledge Ai¾ based on the equivalence relation $\mathcal{R}$ on formulae set F(S) and the corresponding equivalence classification is presented, obtains another algorithm of approximate reasoning -- the IV-type $R$---algorithm.
- –
- Equivalence Knowledge Mass and Approximate Reasoning in -Logic (I).–
This paper presents another matching scheme between the input A ∗ and the knowledge A → B based on the equivalence relation R on formulae set F (S) and the corresponding equivalence classification F(S)/R = {(A)R |A ∈F (S)}
- –
- –
- Ranking with semi-supervised distance metric learning and its application to housing potential estimation–
It is shown that the semi-supervised distance metric learning algorithm is efficient to help consultants to refine their consulting work and can be used to visualize high-dimensional data.
- Support vector machine based Chinese dialect identification–
The experimental results show that SVM based classifier can raise the rate of correct identification about 20% and 4% respectively compared with traditional discriminative classifier and Artificial Neural Network.
- –
- Exemplar-Based Human Contour Tracking–
An exemplar-based tracking framework for human contour tracking with main task to generate contours to fill in the gaps in-between in the test sequences, and to supply the dynamics for updating the target contour by fast contour query.
- New Boosting Methodsof GaussianProcessesfor–
This paper develops twoboosting methods of Gaussian Process for regression according to the characteristic of them and finds that these methods are morestable and have less over-fitting problems than other boosting algorithms.
- –
- –
- –
- A New Adaptive Self-Organizing Map–
A new model based on Kohenen SOM is proposed that has integrated a series of evolutionary working mechanisms of neurons that can conquer the problems of instability in competitive learning.
- Multiple face location using motion information–
Two new methods are presented for multiple face location via motion analysis techniques based on motion segmentation and derived from the Hough Transform, which is based on modeling a head outline as a curve consisting of circle segments.
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.