Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks330 from public data
- 4,891
- Generalizing from a Few Examples2,800
A thorough survey to fully understand Few-shot Learning and categorize FSL methods from three perspectives: data, which uses prior knowledge to augment the supervised experience; model, which uses prior knowledge to reduce the size of the hypothesis space; and algorithm, which uses prior knowledge to alter the search for the best hypothesis in the given hypothesis space.
- 936
- 780
- Transfer learning via dimensionality reduction727
A new dimensionality reduction method is proposed to find a latent space, which minimizes the distance between distributions of the data in different domains in a latentspace, which can be treated as a bridge of transferring knowledge from the source domain to the target domain.
- 504
- Combination of images with diverse focuses using the spatial frequency465
An approach that fuses images with diverse focuses by first decomposing the source images into blocks and then combining them by the use of spatial frequency is presented.
- Timeseries Anomaly Detection using Temporal Hierarchical One-Class Network457
The proposed Temporal Hierarchical One-Class (THOC) network is a temporal one-class classification model for timeseries anomaly detection that captures temporal dynamics in multiple scales by using a dilated recurrent neural network with skip connections.
- Multi-Label Learning with Global and Local Label Correlation421
This paper proposes a new multi-label approach GLOCAL dealing with both the full-label and the missing-label cases, exploiting global and local label correlations simultaneously, through learning a latent label representation and optimizing label manifolds.
- Asynchronous Distributed ADMM for Consensus Optimization412
An asynchronous ADMM algorithm is proposed by using two conditions to control the asynchrony: partial barrier and bounded delay and achieves faster convergence than its synchronous counterpart in terms of the wall clock time.
- Improved Nyström low-rank approximation and error analysis355
An error analysis that directly relates the Nyström approximation quality with the encoding powers of the landmark points in summarizing the data is presented, and the resultant error bound suggests a simple and efficient sampling scheme, the k-means clustering algorithm, for NyStröm low-rank approximation.
- Using the discrete wavelet frame transform to merge Landsat TM and SPOT panchromatic images347
Experimental results show that the proposed approach outperforms methods based on the intensity-hue-saturation transform, principal component analysis and discrete wavelet transform in preserving spectral and spatial information, especially in situations where the source images are not perfectly registered.
- Multifocus image fusion using artificial neural networks302
Experimental results show that the proposed method outperforms the discrete wavelet transform based approach, particularly when there is a movement in the objects or misregistration of the source images.
- Mining customer product ratings for personalized marketing272
This paper addresses some issues faced by recommender systems, and study how recent machine learning algorithms, namely the support vector machine and the latent class model, can be used to alleviate these problems.
- 237
- 228
- Simpler core vector machines with enclosing balls225
This paper introduces the enclosing ball (EB) problem where the ball's radius is fixed and thus does not have to be minimized, and develops efficient (1 + e)-approximation algorithms that are simple to implement and do not require any numerical solver.
- Texture classification using the support vector machines214
This paper applies SVMs for texture classification, using translation-invariant features generated from the discrete wavelet frame transform, and uses a fusion scheme based on multiple SVMs, each with a different setting of the kernel parameter.
- Efficient Multi-label Classification with Many Labels196
This paper addresses the problem of computationally inefficient multi-label classification by selecting a small subset of class labels that can approximately span the original label space by an efficient randomized sampling procedure.
- Accelerated Gradient Method for Multi-task Sparse Learning Problem194
An accelerated gradient method based on an ``optimal'' first order black-box method named after Nesterov and provide the convergence rate for smooth convex loss functions that significantly outperforms the most state-of-the-art methods in both convergence speed and learning accuracy.
- 193
- Learning with idealized kernels190
This paper considers the problem of adapting the kernel so that it becomes more similar to the so-called ideal kernel as a distance metric learning problem that searches for a suitable linear transform (feature weighting) in the kernel-induced feature space.
- 188
- Accelerated Gradient Methods for Stochastic Optimization and Online Learning186
The proposed algorithm, SAGE (Stochastic Accelerated GradiEnt), exhibits fast convergence rates on stochastic composite optimization with convex or strongly convex objectives and can be extended for online learning, resulting in a simple algorithm but with the best regret bounds currently known for these problems.
- Multi-Label Classification on Tree- and DAG-Structured Hierarchies185
The proposed algorithm is computationally efficient, easy to implement, does not suffer from the problem of insufficient/skewed training data in classifier training, and can be readily used on large hierarchies.
- 183
- Non-autoregressive Conditional Diffusion Models for Time Series Prediction180
TimeDiff is proposed, a non-autoregressive diffusion model that achieves high-quality time series prediction with the introduction of two novel conditioning mechanisms: future mixup and autoregressive initialization.
- 173
- 165
- The evidence framework applied to support vector machines164
It is shown that training of the support vector machine (SVM) can be interpreted as performing the level 1 inference of MacKay's evidence framework, which allows automatic adjustment of the regularization parameter and the kernel parameter to their near-optimal values.
- 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.
- 159
- Making large-scale nyström approximation possible142
An accurate and scalable Nystrom scheme that first samples a large column subset from the input matrix, but then only performs an approximate SVD on the inner submatrix by using the recent randomized low-rank matrix approximation algorithms.
- Automated Text Categorization Using Support Vector Machine.138
SVM adapts eeciently in dynamic environments that require frequent additions to the document collection, and allows easy incorporation of new documents into an existing trained system.
- A Survey on Time-Series Pre-Trained Models134
This survey provides a comprehensive review of Time-Series Pre-Trained Models (TS-PTMs), aiming to guide the understanding, applying, and studying TS-PTMs.
- Adaptive localization in a dynamic WiFi environment through multi-view learning133
LeManCoR, a system for adapting the mapping function between the signal space and physical location space over different time periods based on Manifold Co-Regularization is described and it is shown that LeMan co-CoR can effectively transfer the knowledge between two time periods without requiring too much new calibration effort.
- 133
- Tighter and convex maximum margin clustering131
A novel convex optimization method, LG-MMC, which maximizes the margin of opposite clusters via “Label Generation”, which is much more scalable than existing convex approaches and tighter than state-of-art convex MMCs.
- Mixture of Cluster-Conditional LoRA Experts for Vision-Language Instruction Tuning128
A novel Mixture of Experts (MoE) architecture designed to activate task-customized model parameters based on instruction clusters is proposed, and a separate universal expert is further incorporated to improve generalization abilities of MoCLE for novel instructions.
- KICGPT: Large Language Model with Knowledge in Context for Knowledge Graph Completion127
KICGPT is proposed, a framework that integrates a large language model (LLM) and a triple-based KGC retriever that alleviates the long-tail problem without incurring additional training overhead.
- Searching to Exploit Memorization Effect in Learning with Noisy Labels124
This paper designs a domain-specific search space based on general patterns of the memorization effect and proposes a novel Newton algorithm to solve the bi-level optimization problem ef ficiently and provides theoretical analysis of the algorithm, which ensures a good approximation to critical points.
- Fast Stochastic Alternating Direction Method of Multipliers123
A new stochastic alternating direction method of multipliers (ADMM) algorithm, which incrementally approximates the full gradient in the linearized ADMM formulation, which improves the convergence rate on convex problems from O(1/√T) to O( 1/T ), where T is the number of iterations.
- 117
- RouterDC: Query-Based Router by Dual Contrastive Learning for Assembling Large Language Models113
The RouterDC model consists of an encoder and LLM embeddings, and the proposed two contrastive learning losses to train the RouterDC model are proposed, which is effective in assembling LLMs and largely outperforms individual top-performing LLMs as well as existing routing methods.
- Semi-supervised learning using label mean113
Two versions of the meanS3VM, which work by maximizing the margin between the label means, are proposed, which are based on multiple kernel learning, while the second one is based on alternating optimization.
- Text detection in images using sparse representation with discriminative dictionaries109
A classification-based algorithm for text detection using a sparse representation with discriminative dictionaries that can effectively detect texts of various sizes, fonts and colors from images and videos.
- 105
- 104
- Large-Scale Low-Rank Matrix Learning with Nonconvex Regularizers103
This paper develops a fast proximal algorithm and its accelerated variant with inexact proximal step, and shows the proposed algorithm can be parallelized, and the resultant algorithm achieves nearly linear speedup w.r.t. the number of threads.
- Convex and scalable weakly labeled SVMs103
This paper focuses on SVMs and proposes the WELLSVM via a novel label generation strategy, which leads to a convex relaxation of the original MIP, which is at least as tight as existing convex Semi-Definite Programming (SDP) relaxations.
- A Convex Method for Locating Regions of Interest with Multi-instance Learning102
Two convex optimization methods which maximize the margin of concepts via key instance generation at the instance-level and bag-level, respectively are proposed which can effectively locate ROIs and achieve performances competitive with state-of-the-art algorithms on benchmark data sets.
- Linear dependency between Σ and the input noise in Σ-support vector regression101
The resultant predicted choice of /spl epsi/ is much closer to the experimentally observed optimal value, while again demonstrating a linear trend with the input noise.
- Prototype vector machine for large scale semi-supervised learning100
This work proposed the prototype vector machine (PVM), a highly scalable, graph-based algorithm for large-scale semi-supervised learning (SSL), with key innovation is the use of "prototypes vectors" for efficient approximation on both the graph- based regularizer and model representation.
- 97
- 95
- 94
- Cost-Sensitive Semi-Supervised Support Vector Machine94
It is shown that the CS4VM, when given the label means of the unlabeled data, closely approximates the supervised cost-sensitive SVM that has access to the ground-truth labels of all the unl labeled data.
- Multilabel Classification with Label Correlations and Missing Labels92
A probabilistic model that can automatically learn and exploit multilabel correlations by integrating out the missing information and provides a disciplined approach to the handling of missing labels is proposed.
- 92
- 91
- Distance metric learning with kernels89
A feature weighting method that works in both the input space and the kernel-induced feature space that assumes only the availability of similari ty (dissimilarity) information, and the number of parameters in the transformation does not depend on thenumber of features.
- Efficient Learning of Timeseries Shapelets88
The shapelet discovery task is reformulated as a numerical optimization problem and the shapelet positions are learned by combining the generalized eigenvector method and fused lasso regularizer to encourage a sparse and blocky solution.
- Large-Scale Sparsified Manifold Regularization87
This paper integrates manifold regularization with the core vector machine and produces sparse solutions with low time and space complexities by using a sparsified manifold regularizer and formulating as a center-constrained minimum enclosing ball problem.
- A regularization framework for multiple-instance learning87
A more complete regularization framework for MI learning is provided by allowing the use of different loss functions between the outputs of a bag and its associated instances by using the constrained concave-convex procedure.
- 81
- 80
- Covariate Shift in Hilbert Space: A Solution via Sorrogate Kernels79
This work proposes to match data distributions in the Hilbert space, which, given a pre-defined empirical kernel map, can be formulated as aligning kernel matrices across domains, and introduces the novel concept of surrogate kernel based on the Mercer's theorem.
- Density-Weighted Nyström Method for Computing Large Kernel Eigensystems77
An efficient algorithm is proposed to enforce such weighting in practice, which has the same complexity as the original Nystrm method and hence is notably cheaper than several other alternatives.
- Efficient Inexact Proximal Gradient Algorithm for Nonconvex Problems74
Experiments on image inpainting and matrix completion problems demonstrate that the proposed algorithm has comparable performance as the state-of-the-art, but is much faster.
- Transferring localization models across space72
A novel approach to transferring the learning model trained on data from one area of a building to another by solving an optimization problem based on manifold learning techniques and can significantly reduce the amount of labeled data which are required for building the localization model.
- Time Series Anomaly Detection with Multiresolution Ensemble Decoding69
A simple yet efficient recurrent network ensemble called Recurrent Autoencoder with Multiresolution Ensemble Decoding (RAMED), which outperforms recent strong baselines on time series anomaly detection by using decoders with different decoding lengths and a new coarse-to-fine fusion mechanism.
- 68
- 67
- 66
- Fast-and-Light Stochastic ADMM66
This paper proposes an integration of ADMM with the method of stochastic variance reduced gradient (SVRG), which retains the fast convergence benefits of SAG-adMM and SDCA-ADMM, but is more advantageous in that its storage requirement is very low, even independent of the sample size n.
- Marginalized multi-instance kernels66
This paper addresses this instance label ambiguity by using the method of marginalized kernels, which first assumes that all the instance labels are available and defines a label-dependent kernel on the instances, and which outperforms a number of traditional MI learning methods.
- EMOVA: Empowering Language Models to See, Hear and Speak with Vivid Emotions64
This work proposes the EMOVA (EMotionally Omni-present Voice Assistant), to enable Large Language Models with end-to-end speech abilities while maintaining the leading vision-language performance and surprisingly notices that omni-modal alignment can further enhance vision-language and speech abilities compared with the bi-modal aligned counterparts.
- Forward-Backward Reasoning in Large Language Models for Mathematical Verification63
This work proposes FOBAR to combine FOrward and BAckward Reasoning for verification for verification, and shows that FOBAR outperforms Self-Consistency, which uses forward reasoning alone, demonstrating that combining forward and forward reasoning is better.
- 63
- Learning to predict from crowdsourced data61
A novel mixture model is employed for worker annotations, which learns a prediction model directly from samples to labels for efficient out-of-sample testing.
- Dissimilarity learning for nominal data60
This paper proposes the use of adaptive dissimilarity matrices for measuring the dissimilarities between nominal values and shows that this approach leads to better classification performance and allows easier interpretation of (dis)similarity between different nominal values.
- 59
- Parametric distance metric learning with label information57
A parametric method for metric learning based on class label information is proposed, which performs parametric learning to find a regression mapping from the input space to a feature space, such that the dissimilarity between patterns in the input space is approximated by the Euclidean distance between points in the feature space.
- Efficient Sparse Low-Rank Tensor Completion Using the Frank-Wolfe Algorithm52
This paper proposes a time and space-efficient low-rank tensor completion algorithm by using the scaled latent nuclear norm for regularization and the Frank-Wolfe algorithm for optimization.
- SVDD-Based Pattern Denoising52
This paper combines the geodesic projection to the spherical decision boundary resulting from the SVDD, together with solving the preimage problem, and proposes a new method for pattern denoising.
- An Adaptive Policy to Employ Sharpness-Aware Minimization51
This paper designs an adaptive policy to employ SAM based on the loss landscape geometry and proposes two efficient algorithms, AE-SAM and AE-LookSAM, theoretically showing that AE- SAM has the same convergence rate as SAM.
- 51
- End-to-end privacy control in service outsourcing of human intensive processes: A multi-layered Web service integration approach50
This work proposes a layered architecture and a development methodology for enforcing end-to-end privacy control policies of enterprises over the export of personal information and illustrates how Web services, augmented with updated privacy facilities, can provide a suitable interoperation platform for service outsourcing.
- Core Vector Regression for very large regression problems49
The recently proposed Core Vector Machine algorithm is extended to the regression setting by generalizing the underlying minimum enclosing ball problem, and the resultant Core Vector Regression (CVR) algorithm is developed.
- 49
- Scaling Up Graph-Based Semisupervised Learning via Prototype Vector Machines48
This paper proposes to scale up graph-based graph-based semisupervised learning using a set of sparse prototypes derived from the data that serve as a small set of data representatives, which can be used to approximate the graph-based regularizer and to control model complexity.
- 48
- 48
- Normalization Helps Training of Quantized LSTM47
It is shown theoretically that training a quantized LSTM is difficult because quantization makes the exploding gradient problem more severe, and that the popularly used weight/layer/batch normalization schemes can help stabilize the gradient magnitude in training quantization LSTMs.
- Fast Low-Rank Matrix Learning with Nonconvex Regularization47
This paper shows that for many commonly-used nonconvex low-rank regularizers, a cutoff can be derived to automatically threshold the singular values obtained from the proximal operator, which allows the use of power method to approximate the SVD efficiently.
- Linear Dependency between ε and the Input Noise in ε-Support Vector Regression47
This paper attempts to better explain the experimental results there, by analyzing a toy problem with a closer setting to the e-SVR and predicted the value of the theoretically optimal e is much closer to the experimentally observed value, while still demonstrating a linear trend with the data noise.
- Efficient Neural Interaction Function Search for Collaborative Filtering46
Experimental results demonstrate that the proposed method can be much more efficient than popular AutoML approaches, can obtain much better prediction performance than state-of-the-art CF approaches, and can discover distinct IFCs for different data sets and tasks.
- Effective Decoding in Graph Auto-Encoder Using Triadic Closure44
This paper incorporates the well-known triadic closure property, which is exhibited in many real-world networks, to the triad decoder, which considers and predicts the three edges involved in a local triad together in any graph-based auto-encoder.
- Very Large SVM Training using Core Vector Machines.44
This paper scales up kernel methods by exploiting the “approximateness” in practical SVM implementations, formulate many kernel methods as equivalent minimum enclosing ball problems in computational geometry, and then obtain provably approximately optimal solutions efficiently with the use of core-sets.
- Scalable Online Convolutional Sparse Coding41
This paper reformulation of the CSC objective so that convolution can be handled easily in the frequency domain, and much smaller history matrices are needed, and uses the alternating direction method of multipliers (ADMMs) to solve the resultant optimization problem.
- 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.
- 41
- Analytical Survey of Learning with Low-Resource Data: From Analysis to Investigation40
This survey employs agnostic active sampling theory within the Probably Approximately Correct (PAC) framework to analyze the generalization error and label complexity associated with learning from low-resource data in both model-agnostic supervised and unsupervised settings.
- Accelerated and Inexact Soft-Impute for Large-Scale Matrix and Tensor Completion40
This paper shows that Soft-Impute can indeed be accelerated without comprising the special “sparse plus low-rank” structure, leading to low iteration complexity and fast convergence rate, and proposes an approximate singular value thresholding scheme based on the power method.
- Surrogate maximization/minimization algorithms and extensions40
The usefulness of SM algorithms is demonstrated by taking logistic regression models, AdaBoost and the log-linear model as examples and devise several SM algorithms, including the standard SM, generalized SM, gradient SM, and quadratic SM algorithms.
- Towards Safe Semi-Supervised Learning for Multivariate Performance Measures39
Experimental results show that the proposed UMVP (safe semi-supervised learning for MultiVariate Performance measure) method can effectively improve the safeness of SSL under multiple multivariate performance measures.
- 39
- Alignment of Diffusion Models: Fundamentals, Challenges, and Future38
This work mainly reviews alignment of diffusion models, covering advancements in fundamentals of alignment, alignment techniques of diffusion models, preference benchmarks, and evaluation for diffusion models.
- 38
- Collaborative Filtering with Social Local Models38
SLOMA is the first work to incorporate social connections into the local low-rank framework by applying social regularization to submatrices factorization, denoted as SLOMA++ and experimental results from two real-world datasets demonstrate the superiority of the proposed models over LLORMA and MF.
- Accelerated inexact soft-impute for fast large-scale matrix completion38
This paper shows that Soft-Impute can indeed be accelerated without compromising the "sparse plus low-rank" structure and proposes an approximate singular value thresholding scheme based on the power method that enjoys the fast O(1=T2) convergence rate of accelerated proximal gradient algorithms.
- 37
- Dual-balancing for multi-task learning36
DB-MTL achieves loss-scale balancing by performing logarithm transformation on each task loss, and rescales gradient magnitudes by normalizing all task gradients to comparable magnitudes using the maximum gradient norm.
- Bilinear Scoring Function Search for Knowledge Graph Learning36
This paper searches bilinear scoring functions for different KG tasks through the AutoML techniques and proposes a progressive algorithm (AutoBLM) and an evolutionary algorithm ( autoBLM+), which are further accelerated by filter and predictor to deal with the domain-specific properties for KG learning.
- 36
- Efficient kernel feature extraction for massive data sets36
Comparisons with the original MMDA, KPCA, and KFD on a number of large data sets show that the proposed feature extractor can improve classification accuracy, and is also faster than these kernel-based methods by more than an order of magnitude.
- 36
- Accurate and low-cost location estimation using kernels35
A novel approach to building a mapping between the signalvector space and the physical location space using kernel canonical correlation analysis (KCCA), which requires much less human labor than previous methods.
- Gradient-Based Multi-Objective Deep Learning: Algorithms, Theories, Applications, and Beyond34
A comprehensive survey of gradient-based techniques for multi-objective deep learning is provided, systematically categorizing existing algorithms based on their outputs: methods that find a single, well-balanced solution, methods that generate a finite set of diverse Pareto-optimal solutions, and methods that learn a continuous Pareto set of solutions.
- Illumination Controllable Dehazing Network based on Unsupervised Retinex Embedding34
A multi-output dehazing network by introducing illumination controllable ability, called IC-Dehazing, which can change the illumination intensity by adjusting the factor of the illumination controllable module, which is realized based on the interpretable Retinex model.
- 34
- 34
- Gradient Descent with Proximal Average for Nonconvex and Composite Regularization33
A novel proximal gradient descent method for optimization with a wide class of nonconvex and composite regularizers by average the solutions from the proximal problems of the constituent regularizers is proposed.
- Multi-objective Neural Architecture Search via Predictive Network Performance Optimization32
Inspired by the nature of the graph structure of a neural network, BOGCN-NAS is proposed, a NAS algorithm using Bayesian Optimization with Graph Convolutional Network (GCN) predictor to adaptively discover and incorporate nodes structure to approximate the performance of the architecture.
- 32
- 32
- Gene Feature Extraction Using T-Test Statistics and Kernel Partial Least Squares31
Experimental results demonstrate that using the T-test method and kernel partial least squares together can significantly boost classification accuracy, and this simple combination can obtain state-of-the-art performance on all three datasets.
- 30
- 30
- 29
- Surrogate maximization/minimization algorithms for AdaBoost and the logistic regression model29
This paper solves the boosting problem by proposing SM algorithms for the corresponding optimization problem and derives an SM algorithm that can be shown to be identical to the algorithm proposed by Collins et al. (2002) based on Bregman distance.
- 28
- 28
- Accelerated Stochastic Gradient Method for Composite Regularization27
On both general convex and strongly convex problems, the resultant approximation errors reduce at a faster rate than methods based on stochastic smoothing and ADMM, which is also veried experimentally on a number of synthetic and real-world data sets.
- Advances in Neural Networks - ISNN 201027
Introducing a new hobby for other people may inspire them to join with you, and one that the authors will refer to break the boredom in reading is choosing advances in neural networks as the reading material.
- 27
- Implicit Concept Removal of Diffusion Models26
The Geom-Erasing effectively mitigates the generation of implicit concepts, achieving the state-of-the-art results on the Inappropriate Image Prompts (I2P) and the authors' challenging Implicit Concept Dataset (ICD) benchmarks.
- Accelerated Convergence Using Dynamic Mean Shift26
Experiments on color image segmentation show that dynamic mean shift produces comparable results as the standard mean shift algorithm, but can significantly reduce the number of iterations for convergence and takes much less time.
- Incremental eigen decomposition26
By using a method for updating singular value decompositions, this paper proposes a procedure to obtain an approximate eigen decomposition by processing the samples in chunks or sequentially and experimental results show that this form of updating can still achieve comparable performance as the batch version.
- 26
- Asynchronous Distributed Semi-Stochastic Gradient Optimization25
This paper proposes a fast distributed asynchronous SGD-based algorithm with variance reduction that outperforms state-of-the-art distributed asynchronous algorithms in terms of both wall clock time and solution quality.
- 25
- DocPruner: A Storage-Efficient Framework for Multi-Vector Visual Document Retrieval via Adaptive Patch-Level Embedding Pruning24
DocPruner is introduced, the first framework to employ adaptive patch-level embedding pruning for VDR to effectively reduce the storage overhead, and enables a significant 50-60% reduction in storage for leading multi-vector VDR models with negligible degradation in document retrieval performance.
- Effective Structured Prompting by Meta-Learning and Representative Verbalizer24
A prompt pool is used to extract more task knowledge and construct instance-dependent prompts via attention and a novel soft verbalizer (RepVerb) is proposed which constructs label embedding from feature embeddings directly and is parameter-efficient as only the pool is required to be tuned.
- Feedback Pyramid Attention Networks for Single Image Super-Resolution24
A novel feedback connection structure is developed to enhance low-level feature expression with high-level information and introduce a pyramid non-local structure to model global contextual information in different scales and improve the discriminative representation of the network.
- Query Rewriting in TaoBao Search24
This work first alleviate the representation degeneration problem during the query retrieval stage by using an unsupervised contrastive loss, and then proposes an interaction-aware matching method to find the beneficial and incremental candidates, thus improving the quality and relevance of candidate queries.
- Efficient Nonconvex Regularized Tensor Completion with Structure-aware Proximal Iterations23
An efficient solver that avoids expensive tensor folding and unfolding is developed based on the proximal average algorithm and is also more accurate than existing approaches.
- Fast Distributed Asynchronous SGD with Variance Reduction.23
This paper proposes a distributed asynchronous SGD-based algorithm with variance reduction that outperforms state-of-the-art distributed asynchronous algorithms in terms of both wall clock time and solution quality.
- Model-based transductive learning of the kernel matrix23
This paper addresses the problem of transductive learning of the kernel matrix from a probabilistic perspective and devise an expectation-maximization (EM) algorithm to infer the missing data, parameter matrix and feature dimensionality in a maximum a posteriori (MAP) manner.
- A Scalable, Adaptive and Sound Nonconvex Regularizer for Low-rank Matrix Learning22
This paper proposes a new nonconvex low-rank regularizer called ”nuclear norm minus Frobenius norm” regularizer, which is scalable, adaptive and sound, and shows it provably holds the adaptive shrinkage property.
- 22
- Locally adaptive classification piloted by uncertainty22
This paper proposes a responsibility mixture model that uses the uncertainty associated with the classification at each training sample to place the classifiers based on the data distribution only, and demonstrates its superiority over traditional algorithms.
- Fooling the Image Dehazing Models by First Order Gradient21
A group of attack methods based on first order gradient to verify the robustness of the existing dehazing algorithms are designed, which are predicted dehazed image attack, hazy layer mask attack, haze-free image attack and haze-preserved attack.
- Greedy learning of generalized low-rank models21
A more flexible greedy algorithm for generalized low-rank models whose optimization objective can be smooth or nonsmooth, general convex or strongly convex, and low per-iteration time complexity and fast convergence rate is developed.
- Side Information Fusion for Recommender Systems over Heterogeneous Information Network20
This work applies metagraph to similarity computation and solves the information fusion problem with a “matrix factorization (MF) + factorization machine (FM)” framework and proposes an end-to-end method, hierarchical attention fusing, to fuse metagraph-based similarities for the final recommendation.
- Bayesian inference for transductive learning of kernel matrix using the Tanner-Wong data augmentation algorithm20
By regarding the transductive learning of the kernel matrix as a missing data problem, a Bayesian hierarchical model is proposed and the Tanner-Wong data augmentation algorithm for making inference on the model is devised.
- Exploring the Coordination of Frequency and Attention in Masked Image Modeling19
This work proposes the Frequency & Attention-driven Masking and Throwing Strategy (FAMT), which can detect semantic patches and reduce the number of training patches to boost model performance and training efficiency simultaneously and introduces a patch throwing strategy based on the derived sampling weights to reduce the training cost.
- CFVNet: An End-to-End Cancelable Finger Vein Network for Recognition19
This work proposes an end-to-end cancelable finger vein network (CFVNet), which can be used to design an secure finger vein recognition system and integrates preprocessing and template protection into an integrated deep learning model.
- 19
- 19
- 19
- Block-quantized kernel matrix for fast spectral embedding19
An efficient approach to solve the eigendecomposition of the kernel matrix W by approximate W with W that is composed of m2 constant blocks, which scales more favorably than state-of-the-art low rank approximation and sampling based approaches and the approximation quality can be controlled conveniently.
- 18
- Bayesian inference on principal component analysis using reversible jump Markov chain Monte Carlo18
This work presents a hierarchical model for probabilistic PCA and constructs a Bayesian inference method for this model using reversible jump Markov chain Monte Carlo (MCMC), capable of automatically determining the number of principal components and estimating the parameters simultaneously under the same disciplined framework.
- Eigenvoice speaker adaptation via composite kernel PCA17
Kernel eigenvoice speaker adaptation, a nonlinear PCA method based on composite kernels, can reduce the word error rate by 28-33% while the standard eigen Voice approach can only match the performance of the speaker-independent model.
- 17
- Integrating the evidence framework and the support vector machine17
It is shown that training of the support vector machine can be interpreted as performing the level 1 inference of MacKay’s evidence framework, and that levels 2 and 3 can also be applied to SVM, allowing automatic adjustment of the regularization parameter and the kernel parameter.
- Privacy-Preserving Stacking with Application to Cross-organizational Diabetes Prediction16
This paper proposes to enhance privacy-preserving logistic regression by stacking, and shows that this can be done either by sample-based or feature-based partitioning, and proves that when privacy-budgets are the same, feature- based partitioning requires fewer samples than sample- based one, and thus likely has better empirical performance.
- Diversified SVM ensembles for large data sets16
Comparisons with the MMDA ensemble along with bagging show that the proposed diversified CVM ensemble can improve classification performance, and is also faster than the original MMDA algorithm by more than an order of magnitude.
- 16
- 15
- Efficient Pareto Manifold Learning with Low-Rank Structure15
This work proposes a novel approach that integrates a main network with several low-rank matrices to efficiently learn the Pareto manifold and significantly reduces the number of parameters and facilitates the extraction of shared features.
- Improving Fast Adversarial Training via Self-Knowledge Guidance14
The Self-Knowledge Guided FAT is formed, leveraging naturally generated knowledge during training to enhance the adversarial robustness without compromising training efficiency, and extensive experiments demonstrate that the SKG-FAT improves the robustness and preserves competitive clean accuracy, outperforming the state-of theart methods.
- 14
- Nonparametric Iterative Machine Teaching14
This paper proposes both random and greedy functional teaching algorithms for Iterative Machine Teaching, which aims to teach nonparametric target models to learners in an iterative fashion and casts NIMT as a functional optimization problem in the function space.
- Efficient Sample-based Neural Architecture Search with Learnable Predictor14
BONAS (Bayesian Optimized Neural Architecture Search), a sample-based NAS framework combined with Bayesian Optimization, which applies Evolution Algorithm method as a sampler and Graph Convolutional Network predictor as a surrogate model to adaptively discover and incorporate nodes structure to approximate the performance of the architecture.
- om Low-Rank Approximation and Error Analysis14
An error analysis is presented that directly relates the Nystr approximation quality with the encoding of the landmark points in summarizing the data and suggests a sim- ple and efficient sampling scheme, the k-means clustering algorithm, for Nystr low-rank approximation quality.
- Applying the Bayesian Evidence Framework to ν-Support Vector Regression14
It is shown that ν-SVR training implies a prior on the size of the Ɛ-tube that is dependent on the number of training patterns, and this prior has properties that are in line with the error-regulating behavior of ν.
- 13
- Leveraging per Image-Token Consistency for Vision-Language Pre-training13
Extensive experiments show that the combination of the EPIC method and state-of-the-art pre-training approaches, including ViLT, ALBEF, METER, and X-VLM, leads to significant improvements on downstream tasks.
- 13
- Online Convolutional Sparse Coding.13
This paper proposes online CSC (OCSC), which processes each sample only once and updates the dictionaries with the help of low-dimensional history matrices recording statistics about past samples, which reduces the memory and computational cost per iteration and shows that convergence to some stationary point of the CSC problem is still guaranteed by the OCSC algorithm.
- 13
- 13
- 13
- LineMVGNN: Anti-Money Laundering with Line-Graph-Assisted Multi-View Graph Neural Networks12
The results show that the proposed LineMVGNN method outperforms state-of-the-art methods, reflecting the effectiveness of money laundering detection with line-graph-assisted multi-view graph learning.
- 12
- Searching a High Performance Feature Extractor for Text Recognition Network12
This article designs a domain-specific search space by exploring principles for having good feature extractors in a huge and complexly structured space and proposes a two-stage algorithm to effectively search in the space.
- Advances in Recommender Systems12
This tutorial focuses on how automated machine learning (AutoML) techniques can benefit the design and usage of recommendation systems and develops a full scope describing what can be automated for recommendation systems.
- 12
- 12
- Speedup of kernel eigenvoice speaker adaptation by embedded kernel PCA12
The two steps of finding the implicit adapted model and its approximate pre-image are integrated by embedding the kernel PCA procedure in the new embedded kernel eigenvoice (eKEV) speaker adaptation method.
- Learning the Relation Between Similarity Loss and Clustering Loss in Self-Supervised Learning11
This paper attempts to leverage the similarity between two distinct images to boost representation in self-supervised learning, and analyzes the relation between similarity loss and feature-level cross-entropy loss.
- 11
- Positive-Unlabeled Node Classification with Structure-aware Graph Learning10
A distance-aware PU loss that uses homophily in graphs to introduce more accurate supervision and a regularizer to align the model with graph structure is proposed.
- 10
- Dropout’s Dream Land: Generalization from Learned Simulators to Reality10
The experimental results show that Dropout's Dream Land is an effective technique to bridge the reality gap between dream environments and reality and a general approach to improve a controller's ability to transfer from a neural network dream environment to reality at little additional cost is presented.
- 10
- Revitalizing Canonical Pre-Alignment for Irregular Multivariate Time Series Forecasting9
KAFNet is proposed, a compact architecture grounded in CPA for IMTS forecasting that couples a Pre-Convolution module for sequence smoothing and sparsity mitigation, a Temporal Kernel Aggregation module for learnable compression and modeling of intra-series irregularity, and Frequency Linear Attention blocks for low-cost inter-series correlation modeling in the frequency domain.
- CodeGraph: Enhancing Graph Reasoning of LLMs with Code9
Compared to the existing methods, CodeGraph demonstrates strong performance on arithmetic problems in graph tasks and offers a more controllable and interpretable approach to the reasoning process.
- 9
- Simple randomized algorithms for online learning with kernels9
This paper proposes two simple and effective stochastic strategies for controlling the budget of the support set in online learning with kernels that have an expected regret that is sublinear in the horizon.
- 9
- Mining customer preference ratings for product recommendation using the support vector machine and the latent class model9
Some problems faced by recommender systems are addressed, and how machine learning techniques, namely the support vector machine and the latent class model, can be used to alleviated them are studied.
- Multi-Order Wavelet Derivative Transform for Deep Time Series Forecasting8
The multi-order Wavelet Derivative Transform (WDT) grounded in the WT is introduced, enabling the extraction of time-aware patterns spanning both the overall trend and subtle fluctuations, and achieves state-of-the-art forecasting accuracy while retaining high computational efficiency.
- Direct Alignment of Language Models via Quality-Aware Self-Refinement8
This work investigates the use of intrinsic knowledge within the on-the-fly fine-tuning LLM to obtain relative qualities and help to refine the loss function, and designs a refinement function to estimate the quality of both the positive and negative responses.
- 8
- 8
- 8
- ColorVein: Colorful Cancelable Vein Biometrics7
An innovative cancelable vein biometric generation scheme, ColorVein, which allows users/administrators to define a controllable pseudo-random color space for grayscale vein images by editing the position, number, and color of hint points, thereby generating protected cancelable templates.
- Automated Dominative Subspace Mining for Efficient Neural Architecture Search7
A novel Neural Architecture Search method via Dominative Subspace Mining (DSM-NAS) that finds promising architectures in automatically mined subspaces and reduces the search cost but also discovers better architectures than state-of-the-art methods in various benchmark search spaces.
- 7
- AlignVE: Visual Entailment Recognition Based on Alignment Relations7
A new architecture called AlignVE is proposed to solve the visual entailment problem with a relation interaction method that models the relation between the premise and hypothesis as an alignment matrix.
- 7
- 7
- Multi-Label learning in the independent label sub-spaces7
The idea in this work is to divide the whole label space to some independent small groups which leads to independent learning and prediction for each small group in the main space, rather than transforming to the compressed space.
- 7
- Fusing Images with Multiple Focuses Using Support Vector Machines7
Experimental results show that the proposed method outperforms the conventional approach based on the discrete wavelet transform and maximum selection rule, particularly when there is slight camera/object movement or mis-registration of the source images.
- 6
- Learning with Heterogeneous Side Information Fusion for Recommender Systems.6
This work applies metagraph to HIN-based RSs and solves the information fusion problem with a "matrix factorization (MF) + factorization machine (FM)" framework, and experimental results show that the proposed approach is better than FM and other state-of-the-art Hin-based recommendation methods.
- More Efficient Accelerated Proximal Algorithm for Nonconvex Problems6
Experimental results on various machine learning problems demonstrate that the proposed APG algorithm has comparable prediction performance as the state-of-the-art, but is much faster.
- 6
- 6
- Linear Dependency Between and the Input Noise in-Support Vector Regression6
This paper analyzes a toy problem with a closer setting to the SVR algorithm, and predicts the resultant predicted choice of optimal is much closer to the experimentally observed value, while still demonstrating a linear trend with the data noise.
- Towards end-to-end privacy control in the outsourcing of marketing activities6
A layered architecture and a development methodology for end-to-end privacy control over the export of each individual customer's records through a Web services platform, according to the corresponding enterprise's privacy control policies are proposed.
- Eigenvoice Speaker Adaptation via Composite Kernel Principal Component Analysis6
Kernel eigenvoice speaker adaptation, a nonlinear PCA method based on composite kernels, can reduce the word error rate by 28–33% while the standard eigen Voice approach can only match the performance of the speaker-independent model.
- 5
- 5
- DynamicMind: A Tri-Mode Thinking System for Large Language Models5
The framework's core innovations include expanding the established dual-process framework of fast and slow thinking into a tri-mode thinking system involving a normal thinking mode to preserve the intrinsic capabilities of LLM and proposing the Thinking Density metric, which aligns computational resource allocation with problem complexity.
- BYOM: Building Your Own Multi-Task Model For Free5
The proposed PERU-FFT and PERU-LoRA outperform existing merging models method by a large margin and achieve comparable performance to using a fine-tuned model per task.
- 5
- Efficient Low-Rank Semidefinite Programming With Robust Loss Functions5
The crux is on constructing a good optimization surrogate, and it is shown that this surrogate can be efficiently obtained by the alternating direction method of multipliers (ADMM), which is empirically efficient and theoretically guaranteed to converge to a critical point.
- 5
- 5
- 5
- 5
- 5
- Structural, Syntactic, and Statistical Pattern Recognition: Joint IAPR International Workshop, SSPR & SPR 2008, Orlando, USA, December 4-6, 2008. Proceedings ... Vision, Pattern Recognition, and Graphics)5
The papers are organized in topical sections on graph-based methods, probabilistic and stochastic structural models for PR, image and video analysis, shape analysis, kernel methods, recognition and classification, applications, ensemble methods, feature selection, density estimation and clustering, pattern recognition and applications.
- 5
- Facial image reconstruction by SVDD-based pattern de-noising5
This paper considers the problem of reconstructing facial images from the partially damaged ones, and proposes to use the SVDD-based de-noising for the reconstruction, and obtains the image of the reconstructed face by obtaining the pre-image of the projection, and then further processing with its shape and texture information.
- 5
- 4
- SPARD: Defending Harmful Fine-Tuning Attack via Safety Projection with Relevance-Diversity Data Selection4
SPARD is a defense framework that integrates Safety-Projected Alternating optimization with Relevance-Diversity aware data selection, and employs SPAG, which optimizes alternatively between utility updates and explicit safety projections with a set of safe data to enforce safety constraints.
- GrowCLIP: Data-aware Automatic Model Growing for Large-scale Contrastive Language-Image Pre-training4
GrowCLIP is proposed, a data-driven automatic model growing algorithm for contrastive language-image pre-training with continuous image-text pairs as input that adopts a dynamic growth space and seeks out the optimal architecture at each growth step to adapt to online learning scenarios.
- 4
- 4
- 4
- Sliced Coordinate Analysis for Effective Dimension Reduction and Nonlinear Extensions4
A notion of nonlinear effective dimension reduction is proposed and a nonlinear extension of SIR is developed called kernel SIR (KSIR), which are based on principal component analysis and sliced coordinate analysis, respectively.
- Towards end-to-end privacy control in the outsourcing of marketing activities: A web service integration solution4
A layered architecture and a development methodology for end-to-end privacy control over the export of each individual customer's records through a Web services platform, according to the corresponding enterprise's privacy control policies are proposed.
- 3
- 3
- 3
- Mixture of insighTful Experts (MoTE): The Synergy of Thought Chains and Expert Mixtures in Self-Alignment3
The MoTE approach not only outperforms existing methods in aligning LLMs with human values but also highlights the benefits of using self-generated data, revealing the dual benefits of improved alignment and training efficiency.
- 3
- Aggregation Weighting of Federated Learning via Generalization Bound Estimation3
This paper estimates the upper and lower bounds of the second-order origin moment of the shifted distribution for the current local model, and uses these bounds disagreements as the aggregation proportions for weightings in each communication round.
- 3
- 3
- Incorporating cellular sorting structure for better prediction of protein subcellular locations3
This article explores the interdependences between subcellular locations and incorporates them with support vector machines for prediction of protein sub cellular localisation and shows that three of the hierarchical models outperform the traditional ‘flat’ model in terms of tree loss values.
- Fast and accurate kernel density approximation using a divide-and-conquer approach3
The key novelty presented here is the efficient derivation of the representative density function using concepts from function approximation, such that the expensive kernel density estimator can be easily summarized by a highly compact model with very few basis functions.
- 3
- 3
- 2
- 2
- 2
- 2
- Searching for Interaction Functions in Collaborative Filtering2
This paper first design an expressive search space for SIF by reviewing and generalizing existing CF approaches, and proposes to represent the search space as a structured multi-layer perceptron, and design a stochastic gradient descent algorithm which can simultaneously update both architectures and learning parameters.
- 2
- 2
- 2
- Aggregating Crowdsourced Ordinal Labels via Bayesian Clustering2
This paper proposes a novel Bayesian model that clusters workers and items together using the nonparametric Dirichlet process priors, and proposes a new variational inference procedure that is more accurate than the state-of-the-art, and more robust to sparser labels.
- 2
- 2
- Spectral and Semidefinite Relaxations of the CLUHSIC Algorithm2
This paper shows that the recently proposed discriminative view of clustering, which includes the DIFFRAC and DisKmeans algorithms, can also be unified under the CLUHSIC framework.
- 2
- 2
- RxEval: A Prescription-Level Benchmark for Evaluating LLM Medication Recommendation1
RxEval is a prescription-level benchmark that evaluates LLM prescribing capability by multiple-choice questions: each question presents a detailed patient profile and time-ordered clinical trajectory, requiring selection of specific medication-dose-route triples from real prescriptions and patient-specific distractors generated via reasoning-chain perturbation.
- Bubble, Bubble, AI’s Rumble: Why Global Financial Regulatory Incident Re-Porting Is Our Shield Against Systemic Stumbles1
A regulatory-grade global database that synthesises post-trade reporting frameworks with proven incident documentation models from healthcare and aviation is proposed, empowering regulators with unprecedented cross-jurisdictional oversight, financial institutions with seamless compliance integration, and investors with critical visibility into previously obscured AI-driven vulnerabilities.
- Beyond scaleup: Knowledge‐aware parsimony learning from deep networks1
This paper attempts to address the brute‐force scaleup of training datasets, learnable parameters and computation power in a parsimonious manner, and builds a framework that uses domain‐specific knowledge as “building blocks” to achieve parsimony in model design, training, and interpretation.
- 1
- Harnessing Adaptive Topology Representations for Zero-Shot Graph Question Answering1
The DynamicTRF framework is developed, which aims to improve both the accuracy and conciseness of graph QA, and introduces a new metric, Graph Response Efficiency (GRE), which measures the balance between the performance and the brevity in graph QA.
- Mixup Augmentation with Multiple Interpolations1
Multi-mix outperforms various mixup variants and non-mixup-based baselines in terms of generalization, robustness, and calibration and can better guide the training process than standard mixup.
- Underwater Organism Color Enhancement via Color Code Decomposition, Adaptation and Interpolation1
Quantitative and visual evaluations on popular and challenging benchmark datasets demonstrate the superiority of ColorCode over existing methods in providing diverse, controllable, and color-realistic enhancement results.
- No Place to Hide: Dual Deep Interaction Channel Network for Fake News Detection based on Data Augmentation1
A novel framework for fake news detection from perspectives of semantic, emotion and data enhancement is proposed, which excavates the emotional evolution patterns of news participants during the propagation process and a dual deep interaction channel network of semantic and emotion is designed to obtain a more comprehensive and fine-grained news representation with the consideration of comments.
- 1
- AntNet: Deep Answer Understanding Network for Natural Reverse QA.1
Experimental results indicate that the proposed deep answer understanding network, called AntNet, for reverse QA is significantly better than existing methods and those modified from classical natural language processing (NLP) deep models.
- 1
- 1
- Fast Nonsmooth Regularized Risk Minimization with Continuation1
This paper proposes a continuation algorithm that is applicable to a large class of nonsmooth regularized risk minimization problems, can be flexibly used with a number of existing solvers for the underlying smoothed subproblem, and with convergence results on the whole algorithm rather than just one of its subproblems.
- 1
- 1
- Incremental peA Based Face Recognition1
This paper presents a PCA-based algorithm for face recognition tasks, where the learnill g stops once the training set has been duly processed.
- –
- –
- FVeinSyn: Synthetic Finger Vein Image Generator–
FVeinSyn explicitly decouples synthesis of vascular topology and imaging appearance to mitigate the limitations caused by insufficient training samples, such as inadequate identity diversity and restricted realism.
- –
- –
- –
- –
- Corrupted but Not Broken: Understanding and Mitigating the Negative Impacts of Corrupted Data in Visual Instruction Tuning–
It is found that disabling a small subset of parameters can almost fully restore performance and a corruption-robust training paradigm is introduced that significantly surpasses existing strategies for mitigating the effects of corrupted data.
- Open Your Eyes: Vision Enhances Message Passing Neural Networks in Link Prediction–
This work equips MPNNs with vision structural awareness by proposing an effective framework called Graph Vision Network (GVN), along with a more efficient variant (E-GVN), and demonstrates that with the proposed frameworks, GVN consistently benefits from the vision enhancement across seven link prediction datasets, including challenging large-scale graphs.
- –
- Sandbox: safeguarded multi-label learning through safe optimal transport–
A novel Sandbox mechanism, which iteratively estimates multiple labels using an implicit optimal transport process, constraining label refinements between noisy labels and model predictions within a predefined polytope, effectively limiting error propagation, enhancing stability, and offering greater flexibility is introduced.
- –
- Communication-Efficient and Privacy-Preserving Decentralized Meta-Learning–
This paper proposes LoDMeta (Local Decentralized Meta-learning), which has similar meta-learning accuracy as centralized meta-learning algorithms, but does not require gathering data from each client and is able to better protect data privacy for each client.
- Unrevealed Threats: A Comprehensive Study of the Adversarial Robustness of Underwater Image Enhancement Models–
A first attempt to conduct adversarial attacks on five well-designed UWIE models on three common underwater image benchmark datasets is made and a new evaluation dimension of UWIE models is proposed, revealing the adversarial vulnerability of UWIE models and proposing a new evaluation dimension of UWIE models.
- –
- A Scalable, Adaptive and Sound Nonconvex Regularizer for Low-rank Matrix Completion–
This paper proposes a new nonconvex low-rank regularizer called "nuclear norm minus Frobenius norm" regularizer, which is scalable, adaptive and sound, and discovers its factored form which bypasses the computation of singular values and allows fast optimization by general optimization algorithms.
- Efficient Low-Rank Matrix Learning by Factorizable Nonconvex Regularization.–
ExtensiveLow-rank matrix completion experiments show that the proposed method obtains state-of-the-art recovery performance and is much more efficient than existing low-rank convex / nonconvex regularization and matrix factorization algorithms.
- Policy Prediction Network: Model-Free Behavior Policy with Model-Based Learning in Continuous Action Space–
This paper proposes a novel deep reinforcement learning architecture that was inspired by previous tree structured architectures which were only useable in discrete action spaces and is the first to introduce implicit model-based learning to Policy Gradient algorithms for continuous action space.
- –
- –
- –
- –
- –
- Proposed CFTC cybersecurity testing rules for derivatives market infrastructure–
The CFTC’s recent proposed rulemaking related to cybersecurity testing has been met with approval generally by industry participants, and is part of a broader effort to protect against cybersecurity threats that may affect the functioning of financial markets.
- –
- –
- –
- Accurate Integration of Aerosol Predictions by Smoothing on a Manifold–
This work improves the prediction accuracy by using the observation that the aerosol optical depth varies slowly in the spatial domain by imposing a smoothness constraint by a Gaussian random field on the Earth's surface, which can be considered as a two-dimensional manifold.
- –
- –
- Flexible Nonparametric Kernel Learning with Different Loss Functions–
Experimental results on a number of real-world data sets demonstrate that the proposed method is as accurate as directly solving the SDP, but can be one to two orders of magnitude faster.
- –
- –
- –
- –
- –
- –
- –
- –
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 2026-10-12. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.