Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks160 from public data
- 383
- Active Learning from Crowds356
A probabilistic model for learning from multiple annotators that can also learn the annotator expertise even when their expertise may not be consistently accurate across the task domain is employed.
- 245
- 211
- 198
- Hybrid Heterogeneous Transfer Learning through Deep Learning197
A deep learning approach to learn a feature mapping between cross-domain heterogeneous features as well as a better feature representation for mapped data to reduce the bias issue caused by the cross- domain correspondences.
- Large-scale Robust Deep AUC Maximization: A New Surrogate Loss and Empirical Studies on Medical Image Classification161
This work proposes a new margin-based min-max surrogate loss function for the AUC score that is more robust than the commonly used AUC square loss, while enjoying the same advantage in terms of large-scale stochastic optimization.
- A Unified Analysis of Stochastic Momentum Methods for Deep Learning143
The stability analysis shows that the momentum term can improve the stability of the learned model and hence improve the generalization performance through the uniform stability approach.
- 143
- Deep Adversarial Attention Alignment for Unsupervised Domain Adaptation: The Benefit of Target Expectation Maximization127
This paper proposes an attention alignment scheme on all the target convolutional layers to uncover the knowledge shared by the source domain, and estimates the posterior label distribution of the unlabeled data for target network training.
- 116
- 106
- Robust Semi-Supervised Learning through Label Aggregation58
This paper proposes an efficient RObust Semi-Supervised Ensemble Learning (ROSSEL) method, which generates pseudo-labels for unlabeled data using a set of weak annotators, and combines them to approximate the ground-truth labels to assist semi-supervised learning.
- Dynamic Time Warping Based Adversarial Framework for Time-Series Domain54
A novel framework for the time-series domain referred as Dynamic Time Warping for Adversarial Robustness (DTW-AR) using the dynamic time warping measure is proposed and a principled algorithm justified by theoretical analysis is developed to efficiently create diverse adversarial examples using random alignment paths.
- Graph-based clustering and ranking for diversified image search44
A novel ranking framework, namely cluster-constrained conditional Markov random walk (CCCMRW), which has two key steps: first, cluster images into topics, and then perform Markovrandom walk in an image graph conditioned on constraints of image cluster information is proposed.
- 41
- 41
- Active Learning from Multiple Knowledge Sources40
This work focuses on maximizing the information that an annotator label provides about the true (but unknown) label of the data point and proposes an AL approach for this new scenario motivated by information theoretic principles.
- 39
- On the Faithfulness of Vision Transformer Explanations35
This work introduces Salience-guided Faithfulness Coefficient (SaCo), a novel evaluation metric leveraging essential in-formation of salience distribution and demonstrates that the use of gradient and multi-layer aggregation can markedly enhance the faithfulness of attention-based explanation, shedding light on potential paths for advancing Vision Transformer explainability.
- Probabilistically robust conformal prediction35
A novel adaptive PRCP (aPRCP) algorithm to achieve probabilistically robust coverage is proposed to determine two parallel thresholds, one for data samples and another one for the perturbations on data (aka"quantile-of-quantile"design).
- Search by Ideal Candidates: Next Generation of Talent Search at LinkedIn34
A next generation of talent search paradigm at LinkedIn: Search by Ideal Candidates, which only needs the searcher to input one or several examples of suitable candidates for the position and dynamically adjusts the ranking results balancing between the roles of query and ideal candidates.
- 32
- Multiple graph unsupervised feature selection32
The results support that the multiple graph could be better than the single graph in the unsupervised feature selection, and the overall performance of the proposed method is higher than the other comparisons.
- Discrete Contrastive Diffusion for Cross-Modal Music and Image Generation30
This work introduces a Conditional Discrete Contrastive Diffusion (CDCD) loss and design two contrastive diffusion mechanisms to effectively incorporate it into the denoising process, and forms CDCD by connecting it with the conventional variational objectives.
- FCBench: Cross-Domain Benchmarking of Lossless Compression for Floating-Point Data29
This study extensively evaluates the performance of eight CPU-based and five GPU-based compression methods developed by both database and high-performance computing communities, using 33 real-world datasets assembled in the Floating-point Compressor Benchmark (FCBench).
- 29
- 28
- A Framework of Online Learning with Imbalanced Streaming Data28
This work proposes a general framework for online learning with imbalanced streaming data, where examples are coming sequentially and models are updated accordingly on-the-fly, and presents a rigorous theoretical justification of the proposed framework for the F-measure maximization.
- Few-Shot Text and Image Classification via Analogical Transfer Learning26
This work proposes a novel transfer learning algorithm, which follows an analogical strategy, and transfers both the revised source hypothesis and the target hypothesis (only trained with a few samples) to learn an analogICAL hypothesis.
- 25
- Homotopy Smoothing for Non-Smooth Problems with Lower Complexity than $O(1/\epsilon)$25
This work will show that the proposed HOPS achieved a lower iteration complexity of $\widetilde O(1/\epsilon^{1-\theta})$ with $\theta\in(0,1]$ capturing the local sharpness of the objective function around the optimal solutions.
- Complex Event Detection via Event Oriented Dictionary Learning24
This paper proposes two novel strategies to automatically select semantic meaningful concepts for the event detection task based on both the events-kit text descriptions and the concepts high-level feature descriptions and introduces a novel event oriented dictionary representation based on the selected semantic concepts.
- Neural graph personalized ranking for Top-N Recommendation23
This work proposes neural graph personalized ranking (NGPR) which directly makes use of the user–item interaction information in embedding learning by incorporating the user-item interaction graph in embeding learning.
- 23
- 22
- 22
- Learning Audio-Visual Correlations From Variational Cross-Modal Generation22
This work introduces a novel Variational AutoEncoder framework that consists of Multiple encoders and a Shared decoder (MS-VAE) with an additional Wasserstein distance constraint to tackle the problem of audio-visual correlations.
- Surface Defect Classification of Steel Strip with Few Samples Based on Dual‐Stream Neural Network21
A novel dual‐stream neural network is proposed, which is composed by two streams of sample generation and classification training, and which has the highest classification accuracy of 99.70% among the state‐of‐the‐art classifiers.
- 18
- A Flexible Forecasting Framework for Hierarchical Time Series with Seasonal Patterns: A Case Study of Web Traffic18
The results demonstrate that the proposed flexible hierarchical forecasting framework is superior when applied to hierarchical web traffic prediction problems, and it outperforms alternative time series prediction models in terms of accuracy.
- Revisiting Optical Flow Estimation in 360 Videos17
This paper designs LiteFlowNet360 as a domain adaptation framework from perspective video domain to 360 video domain and adapts it from simple kernel transformation techniques inspired by Kernel Transformer Network to cope with inherent distortion in 360 videos caused by the sphere-to-plane projection.
- Adaptive Sparse Confidence-Weighted Learning for Online Feature Selection17
An Adaptive Sparse Confidence-Weighted (ASCW) algorithm that introduces an `0-norm constraint into the second-order confidence-weighted (CW) learning for feature selection and maintains multiple sparse CW learner with the corresponding cost vector to dynamically select an optimal cost.
- 17
- 16
- Unusual Morphologies of Poly(vinyl alcohol) Thin Films Adsorbed on Poly(dimethylsiloxane) Substrates15
Light plasma oxidation and subsequent PVOH adsorption can be utilized as a means to effectively hydrophilize conventional PDMS substrates, illustrating that the stability and morphology of adsorbed polymer thin films depend on polymer crystallinity as well as substrate physical properties.
- A Music Recommendation Algorithm Based on Hybrid Collaborative Filtering Technique15
A method to fusion item-based and user-based collaborative filtering algorithms into a hybrid one, which considers time as a property and design a method to evaluate the rank of items is proposed.
- Interpolating Video-LLMs: Toward Longer-sequence LMMs in a Training-free Manner14
This paper introduces an alternative video token rearrangement technique that circumvents limitations imposed by the fixed video encoder and alignment projector, and introduces a training-free LLM context window extension method to enable Video-LLMs to understand a correspondingly increased number of visual tokens.
- Scalable maximum margin matrix factorization by active Riemannian subspace search14
This paper forms M3F with a known number of latent factors as the Riemannian optimization problem on a fixed-rank matrix manifold and presents a block-wise nonlinear RiemANNian conjugate gradient method to solve it efficiently.
- Training Robust Deep Models for Time-Series Domain: Novel Algorithms and Theoretical Analysis13
This paper forms a min-max optimization problem over the model parameters by explicitly reasoning about the robustness criteria in terms of additive perturbations to time-series inputs measured by the global alignment kernel (GAK) based distance and proposes a principled stochastic compositional alternating gradient descent ascent (SCAGDA) algorithm.
- 13
- Efficient Semidefinite Spectral Clustering via Lagrange Duality13
Compared with the state-of-the-art spectral clustering methods, the proposed algorithm achieves better clustering performance and is much more efficient and can solve larger-scale SSC problems than those standard interior-point SDP solvers.
- 12
- From Query-By-Keyword to Query-By-Example: LinkedIn Talent Search Approach12
This paper proposes the next generation of talent search at LinkedIn, also referred to as Search By Ideal Candidates, in this system, a searcher provides one or several ideal candidates as the input to hire for a given position and generates a query based on the ideal candidates and uses it to retrieve and rank results.
- The Pitfalls and Promise of Conformal Inference Under Adversarial Attacks11
This study investigates the uncertainty of deep learning models by examining the performance of conformal prediction (CP) in the context of standard adversarial attacks within the adversarial defense community and proposed Uncertainty-Reducing AT (AT-UR) is validates the effectiveness.
- 11
- 10
- Out-of-distribution Detection in Time-series Domain: A Novel Seasonal Ratio Scoring Approach9
Experiments on diverse real-world benchmarks demonstrate that the SRS method is well-suited for time-series OOD detection when compared to baseline methods.
- Exploring Web Images to Enhance Skin Disease Analysis Under A Computer Vision Framework9
A novel transfer learning model is presented, which is able to incorporate external knowledge obtained from the rich and relevant Web images contributed by grassroots, and is easily repeatable by other researchers and extendable to other disease types.
- Path Planning Based on Closed-Form Characterization of Collision-Free Configuration-Spaces for Ellipsoidal Bodies, Obstacles, and Environments9
The approach of “knowing where to look” for path planning in C-spaces with narrow passages is applied and its potential is demonstrated by comparing with the well known sample-based path planning method that does not currently have the ability to take advantage of a priori knowledge of collision-free regions of C-space.
- Causal-DFQ: Causality Guided Data-free Network Quantization8
Inspired by the causal understanding, this work designs a content-style-decoupled generator, synthesizing images conditioned on the relevant and irrelevant factors, and proposes a discrepancy reduction loss to align the intervened distributions of the pre-trained and quantized models.
- 8
- Saying the Unseen: Video Descriptions via Dialog Agents8
This work introduces a novel task that aims to describe a video using the natural language dialog between two agents as a supplementary information source given incomplete visual data and experimentally demonstrates the knowledge transfer process between the two dialog agents and the effectiveness of using thenatural language dialog as a supplement for incomplete implicit visions.
- Recent Advances on Supervised Distance Metric Learning Algorithms8
Recently developed algorithms for supervised distance metric learning based on the partition of pairwise constraints and non-pairwise constraints are reviewed and some representative algorithms are introduced and their respective pros and cons are analyzed.
- 7
- A review of robot path planning algorithms7
Algorithms for path planning are divided into classical algorithms and heuristic algorithms, and the principles, advantages and disadvantages, research progress and so on of each algorithm are introduced.
- 7
- 7
- 7
- 7
- Structured discriminative tensor dictionary learning for unsupervised domain adaptation6
A pseudo-label estimation scheme is developed to combine with discriminant analysis in the algorithm iteration for avoiding the external classifier design and experimental results demonstrates that the proposed method outperforms the state-of-the-art approaches.
- Guest editorial: Image/video understanding and analysis6
This special issue aims to collect recent state-of-the-art achievement on image/video understanding and analysis, especially the work devoted to several new challenges in the field.
- 6
- Learning Sparse Confidence-Weighted Classifier on Very High Dimensional Data6
This paper presents an online-batch CW learning scheme, and then presents a novel paradigm to learn sparse CW classifiers, which essentially identifies feature groups and naturally builds a block diagonal covariance structure, making it very suitable for CW learning over very high-dimensional data.
- 6
- 6
- 5
- 5
- Application of time series to thermal error compensation of machine tools5
Adopt autoregressive distributed lag model of time series analysis, and contract of results among time series model, multiple linear regression model, demonstrate that forecast accuracy can be improved using time seriesmodel, and time seriesAnalysis has a bright future.
- 4
- 4
- 4
- 4
- 4
- Emission Mechanism of Polyatomic Ions Cs2Cl+ and Cs2BO2+ in Thermal Ionization Mass Spectrometry with Various Carbon Materials4
It could be concluded that the emission behavior of the polyatomic ions strongly depends on the microstructure of the carbon materials used, and a surface-induced collision mechanism for formation of such kinds of polyatomic ion in the ionization source of TIMS has been proposed based on the optimized configuration of Cs(2)BO(2)(+) and Cs (2)Cl(+) ions in the gas phase using a molecular dynamics method.
- Finite Element Analysis of Turbulent Flows Using LES and Dynamic Subgrid‐Scale Models in Complex Geometries4
The main objectives of this research are to know more about the structure of turbulent flows, to identify their three-dimensional characteristic, and to study physical effects due to complex fluid flow.
- 4
- Not Optimal but Efficient: A Distinguisher Based on the Kruskal-Wallis Test3
It is shown that in the context of an unknown (and not linear) leakage model there is indeed a distinguisher that outperforms the “optimal” distin-guisher in terms of trace efficiency: it is based on the Kruskal-Wallis test.
- Improve robustness of machine learning via efficient optimization and conformal prediction3
For learning robust objectives, sample‐efficient stochastic optimization algorithms that achieves the optimal (or faster compared to existing algorithms) convergence rates are designed and a framework to analyze the expected prediction set size of CP methods in both standard and adversarial settings is established.
- 3
- 3
- 3
- 3
- 2
- 2
- 2
- 2
- Low Rank Metric Learning via Accelerated Gradient Optimization2
To effectively solve the proposed low rank metric learning problem, an efficient algorithm is developed based on the accelerated gradient method that not only learns an effective metric, but also projects data into an intrinsic low-dimensional subspace.
- Universal Steganalysis Method Based on Multi-domain Features2
A new method of universal steganalysis for BMP images, which had low embedding rate and based on multi-domain features, is proposed and solved the problem of low detection rate for a small amount of data embedding.
- Study on Universal Steganalysis for BMP Images Based on Multi-Domain Features2
Aimed at the most commonly used cover media, digital images, the paper proposed a new method of universal steganalysis method based on multi-domain features to overcome the drawbacks of steganslysis algorithm for specific steganography.
- 2
- Conformalized Uncertainty Regions for Machine Learning-Based Multiple Cognitive Health Measures from Smartwatch Sensor Data1
Compared to other UQ methods, this method constructs smaller prediction regions across all multi-target combinations while maintaining the user-specified coverage, and can be integrated into human-ML collaborative systems to improve diagnostic accuracy and interpretability in sensitive multi-target healthcare tasks.
- Enhance Robustness of Machine Learning with Improved Efficiency1
This talk will survey three pieces of recent research by elaborating the algorithmic idea and theoretical analysis as technical contributions, and elaborate an efficient conformal prediction algorithm that guarantee the robustness of prediction after model is trained.
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- A Method of Universal Steganalysis Using Independence with Image Format1
A new universal steganalysis method based on statistical characteristic of multi-domain features extraction has better detection effects on images with different steganography carriers, and achieves higher independence of image formats and better average detection effects compared with typical universal Steganalysis algorithms at present.
- Periodic Solution for Neutral Type Neural Networks1
The principle aim of this paper is to explore the existence of periodic solution of neural networks model with neutral delay by means of an abstract continuous theorem of k-set contractive operator and some analysis technique.
- 1
- 1
- MSAN-Net: a lightweight multi-scale attention network for transmission line component segmentation and defect detection on edge devices–
Results indicate that MSAN-Net provides an effective balance between segmentation accuracy, robustness, computational efficiency, and practical edge deployment for intelligent transmission line inspection.
- Application of interactive effect evaluation model in improving computer security management and teaching management–
The experimental results show that the interactive effect evaluation model proposed in this paper can play an important role in computer security management and teaching management.
- –
- –
- –
- –
- –
- Time-domain speech separation based on local parallel recursive networks and global fast attention–
This work proposes a novel method that utilizes TasNets to separate mixed speech signals through the use of parallel recurrent neural networks within the block sequence, with an additional recurrent neural network added.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Effects of Domain Partitioning on Parallel Computing of Atrium Airflow Simulation–
The indoor airflow and thermal environment of a ventilated atrium with large space is simulated using the paralleling solver of FLUENT software, and the effect of two domain partitioning methods and various number of subdomains afterdomain partitioning on the simulated results are discussed.
- Continuous Semantic Inspection–
It is hypothesized that with the extra semantic information developers can more easily determine whether the code changes produced the desired effect, and that the approach is feasible for open source projects by enabling a timely repair-compile-review feedback loop.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- A Method of Multidimensional Knowledge Push for Product Designers–
A knowledge active push system was realized through forgotten attenuation, similarity filtering and other key methods, playing a good supporting role to designers and greatly reducing the time required for designing.
- –
- New Method of Image Denoising Based on Fractional Wavelet Transform–
The results of experiment show that the novel methods could effectively remove noise, and maintain information quantity maximally at the same time by adjusting the fractional order p and wavelet scale.
- –
- Cleaning contest robot and its control strategy–
In the fourth robot competition of jiangsu province, the coverage rate of dust area is 80% and the system has good performance, and achieve the design requirement.
- –
- –
- –
- Application Skills of Discoverer-based Statement Composition–
The paper introduced fundamental conceptions, connection methods and basic environment of Discoverer, and gave an example to illustrate basic method and technique of creating custom statements with Discoverers.
- –
- –
- –
- –
- –
- Remote Monitor and Control System for Large Granary Based on CAN Bus–
It presented the remote monitor system for large granary based on the CAN field bus and the design and function for the senior management computer and applied control computer in detail.
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-14. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.