Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks167 from public data
- 154
- IM-IAD: Industrial Image Anomaly Detection Benchmark in Manufacturing141
A uniform IM benchmark is proposed, for the first time, to assess how well IAD algorithms perform, which includes various levels of supervision, learning paradigms, memory usage and inference speed, and efficiency.
- Class-Aware Contrastive Semi-Supervised Learning140
This work proposes a general method named Class-aware Contrastive Semi-Supervised Learning (CCSSL), which is a drop-in helper to improve the pseudo-label quality and enhance the model's robustness in the real-world setting and successfully emphasizes clean label learning and simultaneously reduce noisy label learning.
- Multi-depot green vehicle routing problem with shared transportation resource: Integration of time-dependent speed and piecewise penalty cost140
This paper introduces a transportation resource sharing strategy to address the multi-depot green vehicle routing problem, and incorporates the time-dependency of speed as well as piecewise penalty costs for earliness and tardiness of deliveries.
- Morphological Feature Visualization of Alzheimer’s Disease via Multidirectional Perception GAN131
By introducing a novel multidirectional mapping mechanism into the model, the proposed MP-GAN can capture the salient global features efficiently and clearly delineate the subtle lesions via MR image transformations between the source domain and the predefined target domain.
- Artificial intelligence, physiological genomics, and precision medicine129
Physiological genomic readouts in disease-relevant tissues, combined with advanced AI, can be a powerful approach for precision medicine for common diseases.
- Collaborative two-echelon multicenter vehicle routing optimization based on state–space–time network representation128
An empirical case study of a logistics network in Chongqing suggests that the proposed collaboration mechanism with SST network representation can reduce costs, improve transportation efficiency, and contribute to efficient and sustainable logistics network operations.
- 112
- 103
- 86
- Neural Retrievers are Biased Towards LLM-Generated Content76
A quantitative evaluation of IR models in scenarios where both human-written and LLM-generated texts are involved indicates that neural retrieval models tend to rank LLM-generated documents higher, and proposes a plug-and-play debiased constraint for the optimization objective to mitigate the source bias.
- End-to-End Beam Retrieval for Multi-Hop Question Answering66
Beam Retrieval is introduced, an end-to-end beam retrieval framework for multi-hop QA that helps the authors' supervised reader achieve new state-of-the-art performance and substantially improves the few-shot QA performance of LLMs.
- Multimodal Representations Learning and Adversarial Hypergraph Fusion for Early Alzheimer’s Disease Prediction57
A novel multimodal representation learning and adversarial hypergraph fusion (MRL-AHF) framework for Alzheimer's disease diagnosis using complete trimodal images and provides a possible way to understand the underlying mechanisms of disorder's progression by analyzing the abnormal brain connections.
- The general inner-outer iteration method based on regular splittings for the PageRank problem44
A general inner-outer (GIO) iteration method is presented for solving the PageRank problem and the preconditioner derived from the GIO iteration can be used to accelerate the Krylov subspace methods, such as GMRES method.
- 38
- 37
- 34
- 28
- Neural Modeling of Buying Behaviour for E-Commerce from Clicking Patterns28
This paper provides a direct method in modeling the buying pattern in a clicking session by simply using the time-stamp of the clicks and shows that the result is comparable to using more massive feature engineering that requires session summarizing.
- 28
- 27
- Optimal Synthesis of Mechanisms for Path Generation Using Refined Numerical Representation Based Model and AIS Based Searching Method26
An optimal synthesis method is presented, in which a new path-description method named refined numerical representation is proposed to define the object function of the optimization model for path generation, and then the artificial immune network searching method is introduced to search candidate solutions.
- Infinite Kernel Learning: Generalization Bounds and Algorithms25
A new kernel learning method is proposed based on a novel measure of generalization error, called principal eigenvalue proportion (PEP), which can learn the optimal kernel with sharp generalization bounds over the convex hull of a possibly infinite set of basic kernels.
- 24
- 22
- Decentralized Riemannian Conjugate Gradient Method on the Stiefel Manifold22
DRCGD is the first decentralized Riemannian conjugate gradient algorithm to achieve global convergence over the Stiefel manifold and is free from expensive Riemannian geometric operations such as retractions, exponential maps, and vector transports.
- A novel channelized FB architecture with narrow transition bandwidth based on CEM FRM19
The FRM digital filter design technique and another important technique named complex-exponential modulation (CEM) are exploited and applied to the design of a novel cascaded channelized filter bank to realize selective sub-channel with very narrow transition bandwidth.
- Beyond Prototypes: Semantic Anchor Regularization for Better Representation Learning18
It is revealed that the class prototype is not necessarily to be derived from training features and a novel perspective to use pre-defined class anchors serving as feature centroid to unidirectionally guide feature learning is proposed.
- Preventing Over-Fitting of Cross-Validation with Kernel Stability17
This work proposes a novel version of stability, called kernel stability, which quantifies the perturbation of the kernel matrix with respect to the changes in the training set, and derives a closed form for the estimate of thekernel stability, making the criterion based on the kernel stability computationally efficient.
- 16
- Joint Learning Content and Degradation Aware Feature for Blind Super-Resolution16
The proposed Content and Degradation aware SR Network dubbed CDSR outperforms the existing UDP models and achieves competitive performance on PSNR and SSIM even compared with the state-of-the-art SKP methods.
- Multi-core-, multi-thread-based optimization algorithm for large-scale traveling salesman problem16
Experimental results show that the algorithm proposed can, under the given hardware limitations, take full advantage of multi-core chips and effectively balance the conflict between increasing problem size and computational efficiency and thus acquire satisfactory solutions.
- 15
- 15
- Automated Spectral Kernel Learning15
An efficient learning framework that incorporates the process of finding suitable kernels and model training is proposed that derives data-dependent generalization error bounds and introduces regularization terms to improve the performance.
- Randomized Kernel Selection With Spectra of Multilevel Circulant Matrices15
A randomized kernel selection approach to evaluate and select the kernel with the spectra of the specifically designed multilevel circulant matrices (MCMs), which is statistically sound and computationally efficient.
- 14
- 14
- Evaluating the External and Parametric Knowledge Fusion of Large Language Models13
This investigation reveals that enhancing parametric knowledge within LLMs can significantly bolster their capability for knowledge integration, and identifies persistent challenges in memorizing and eliciting parametric knowledge, and determining parametric knowledge boundaries.
- Linear Kernel Tests via Empirical Likelihood for High-Dimensional Data13
Empirical likelihood ratio statistics for the two-sample test and the goodness-of-fit test are proposed, and the nonparametric Wilks’ theorems for the ELR statistics are proved, which illustrate that the limiting distributions of the proposed ELRStatistics are chi-square distributions.
- 12
- 12
- Federated Learning for Non-IID Data: From Theory to Algorithm11
An excess risk bound for federated learning on non-IID data is presented, which measures the error between the model of federated learning and the optimal centralized model and proposes FedAvgR to improve the performance via additional regularizers to lower the excess risk.
- 11
- Eigenvalues perturbation of integral operator for kernel selection11
This paper introduces new kernel selection criteria based on the eigenvalues perturbation of the integral operator, and establishes the connection between eigen values perturbations and generalization error.
- 11
- Weighted distributed differential privacy ERM: Convex and non-convex10
This paper proposes a weighted distributed differential privacy (WD-DP) empirical risk minimization (ERM) method to train a model in distributed setting, considering different weights of different clients, and guarantees differential privacy by gradient perturbation, adding Gaussian noise, and advances the state-of-the-art on gradient perturbedation method in distributedSetting.
- 10
- 9
- 9
- Reimagining Graph Classification from a Prototype View with Optimal Transport: Algorithm and Theorem8
A novel prototype-based graph classification framework that introduces the Fused Gromov-Wasserstein (FGW) distance in Optimal Transport (OT) as the similarity measure is proposed, which explicitly exploits the structural information on the graph through OT while leading to a more transparent and straightforward classification process.
- 8
- Convolutional spectral kernel learning with generalization guarantees8
A VGG-type 8-layers CSKN is devised, and it outperforms the existing kernel-based networks and popular CNN models on the medium-sized image classification tasks and proves that under suitable initialization, deeper networks lead to tighter error bounds.
- 8
- 8
- Kernel selection with spectral perturbation stability of kernel matrix8
This work introduces a new notion of stability, called the spectral perturbation stability, to study the kernel selection problem and proposes a new kernel selection criterion that can guarantee good generalization properties.
- Revealing sRNA expression profiles of NDM-5-producing CRKP and explore the role of sRNA207 in NDM-5-producing CRKP resistance7
Increased expression of sRNA207 promoted biofilm formation by stabilizing expression of smf-1, which in turn affected the resistance of NDM-5-producing CRKP to carbapenems, and the results suggest that smf-1 is the target gene of sRNA207.
- Continuous speech recognition based on convolutional neural network7
CNNs, which showed success in achieving translation invariance for many image processing tasks, are investigated for continuous speech recognitions in the paper and showed that CNNs outperformed DNNs in the term of the accuracy when CNNs had even smaller model size.
- Extracting viewer interests for automated bookmarking in video-on-demand services7
Experimental results suggest that with the proposed video bookmarking methodology, viewers’ interests within a video can be precisely extracted, and the authors can position bookmarks on the video’s highlight events accurately.
- Vision-Centric 4D Occupancy Forecasting and Planning via Implicit Residual World Models6
IR-WM, an Implicit Residual World Model that focuses on modeling the current state and evolution of the world, establishes a robust bird’s-eye-view representation of the current state from the visual observation and applies an alignment module to calibrate semantic and dynamic misalignments.
- 6
- 6
- Transformers as Intrinsic Optimizers: Forward Inference through the Energy Principle5
The principle of energy as a lens to understand attention-based Transformer models is revisited and a unified energy-based framework is presented which is composed of three key components: the local energy, the global energy, and the employed optimization algorithms.
- 5
- Multi-Modal Molecular Representation Learning via Structure Awareness5
A structure-awareness-based multi-modal self-supervised molecular representation pre-training framework designed to enhance molecular graph representations by leveraging invariant knowledge between molecules and can be seamlessly integrated with any graph-based method and supports multiple molecular data modalities, ensuring both versatility and compatibility.
- 5
- 5
- Semi-supervised vector-valued learning: Improved bounds and algorithms5
A general semi-supervised algorithm for efficiently learning vector-valued functions, incorporating both local Rademacher complexity and Laplacian regularization is proposed, which significantly outperforms the compared methods.
- Learning Rates for Nonconvex Pairwise Learning5
This work develops different uniform convergence of gradients for pairwise learning under different assumptions, based on which they characterize empirical risk minimizer, gradient descent, and stochastic gradient descent.
- Towards Sharp Analysis for Distributed Learning with Random Features5
This paper first extends the optimal rates for distributed learning with random features to the non-attainable case, then reduces the number of required random features via data-dependent generating strategy, and improves the allowed number of partitions with additional unlabeled data.
- Learning Vector-valued Functions with Local Rademacher Complexity.5
A unified learning framework for vector-valued functions based which is solved by proximal gradient descent on the primal is devised, achieving a much better tradeoff between accuracy and efficiency.
- 5
- 5
- 4
- Reliable Federated Disentangling Network for Non-IID Domain Feature4
This paper proposes a novel trusted federated disentangled network, termed TrFedDis, which utilizes feature disentangling to enable the ability to capture the global domain-invariant cross-client representation and preserve local client-specific feature learning.
- 4
- 4
- Distributed Randomized Sketching Kernel Learning4
The proposed DKRR-RS combines sparse randomized sketching, divide-and-conquer and KRR to scale up kernel methods and derives the same learning rate as the exact KRR with greatly reducing computational costs in expectation, at the basic setting, which outperforms previous state of the art solutions.
- 4
- 4
- [Novel type of unperturbed sleep monitoring scheme under pillow based on hidden Markov model].4
Experimental results validate that the proposed noninvasive monitoring system can capture the sleep stages S1-S4 with an accuracy more than 60%, and performs superior to that of the existing sleep staging scheme based on HMM.
- 4
- An optimal pooling strategy applied to high-throughput screening for rare marker-free transformants4
An optimal pooling system, called Accurate and Fast Target Screening, has been developed for high-throughput identifying the rare marker-free transformants and gives unambiguous positive or negative results and should facilitate markers-free transformation.
- 4
- 3
- Adversarial Masked Graph Autoencoders for Improved Graph Representation Learning3
AMGAE -- an Adversarial Masked Graph AutoEncoder enhances the robustness of masked GAEs by integrating an adversarial learning strategy, and reformulates masked GAE learning into a min-max game, which facilitates the learning of robust graph representations.
- 3
- RLPeri: Accelerating Visual Perimetry Test with Reinforcement Learning and Convolutional Feature Extraction3
RLPeri, a reinforcement learning-based approach to optimize visual perimetry testing that results in a 10-20% reduction in examination time while maintaining the accuracy as compared to state-of-the-art methods.
- Hybrid Augmented Automated Graph Contrastive Learning3
An improved joint training strategy is proposed, which can achieve better results than previous works without resorting to any weak label information in the downstream tasks and extensive evaluation of additional work.
- Sharper Utility Bounds for Differentially Private Models3
By normalization, the high probability excess population risk bound under assumptions α-Hölder smooth and Polyak-Łojasiewicz condition can achieve O (√p over n∈), which is the first O (1/n) high probability excess population risk bound w.r.t n for differentially private algorithms under non-smooth conditions.
- 3
- 3
- Automated Hyper-parameter Tuning for Machine Learning Models in Machine Health Prognostics3
This paper considers the use of Bayesian optimization algorithms, which automate an effective choice of HP-config by solving the associated hyperparameter optimization problem.
- Remote Sensing Image Denoising with Iterative Adaptive Wiener Filter3
A novel Wiener filter strategy is proposed, which takes advantage of image self-similarity to estimate filter parameters adaptively and can significantly improve both the subjective and the objective quality of noisy remote sensing images.
- 3
- 3
- IdmGAE: Importance-Inspired Dynamic Masking for Graph Autoencoders2
This study proposes an efficient masking strategy termed importance-inspired dynamic masking to explore diverse node sampling and introduces a dynamic sampling strategy that adapts to nodes of varying importance across different training stages.
- 2
- SUBP: Soft Uniform Block Pruning for 1xN Sparse CNNs Multithreading Acceleration2
The approach tends to repeatedly allow pruned blocks to regrow to the network based on block angular redundancy and importance sampling in a uniform manner throughout the training process, which makes the model less dependent on pre-training, reduces the model redundancy and the risk of pruning the important blocks permanently but also achieves balanced workload.
- Data Heterogeneity Differential Privacy: From Theory to Algorithm2
This paper first gives sharper excess risk bounds of DP stochastic gradient descent (SGD) method, then introduces the influence function (IF), which quantitatively measure the contributions of various training data on the final machine learning model.
- Shape point set matching based on oriented shape context in turbulence-cluttered scene2
The extensive experiments on both synthetic and real data validate that the proposed energy cost model with figural continuity constraint can effectively detect the desired shapes in the complex and highly turbulence-cluttered scenes.
- Wavelet and Fourier Transforms in Health Monitoring of Embedded Structures2
This research is supported by the National Natural Science foundation of China (Grand No.: 51879203).
- A cross-culture validation on the predictors of online review helpfulness2
The study applied Tobit regression analysis to explore the impact of review extremity and review depth on review helpfulness across six different language groups, including English, Finnish, German, Italian, Russian and Japanese.
- The Fault Diagnosis Method for Roller Bearing with Harmonic Wavelet Packet and DT-SVM2
The multi-classification SVM model is designed due to its superior nonlinear mapping capability and the higher accuracy of the proposed method compared with conventional method is illustrated.
- 2
- 2
- Chemical knowledge-informed framework for privacy-aware retrosynthesis learning1
The chemical knowledge-informed framework (CKIF), a privacy-preserving approach for learning retrosynthesis models, is introduced, enabling organisations to collaboratively train retrosynthesis models without sharing raw data.
- 1
- Short term PV power generation prediction based on wavelet transform and LSTM1
This article proposes a WT-LSTM PV short-term power generation prediction method, where a Long Short Term Memory (LSTM) network is used to predict each component, addressing the negative impact of data volatility and randomness on the prediction results caused by a single LSTM prediction algorithm.
- 1
- Non-IID Distributed Learning with Optimal Mixture Weights1
This paper presents a general framework to optimize the mixture weights and shows that this framework has lower expected loss than the uniform mixture weights framework theoretically and provides strong generalization guarantee for this framework.
- 1
- 1
- 1
- Optimization of Regional City Economic Gravity Network Based on Optimal Tree Theory1
The research method was applied to build and optimize the economic gravity network of the cities in the Yangtze River Delta, based on the key indices of these cities in 2018, and lay a solid theoretic basis for regional planning and national development.
- 1
- 1
- 1
- 1
- A Euclidean Projection Algorithm with Its Application to Image Restoration–
Based on alternating direction method of multipliers and the projection of a matrix onto the given convex set, an efficient image restoration method is provided.
- –
- –
- –
- –
- –
- Towards Sharper Risk Bounds for Minimax Problems–
This paper obtains a sharper high probability generalization error bound for nonconvex-strongly-concave (NC-SC) stochastic minimax problems and analyzes some popular algorithms such as empirical saddle point (ESP), gradient descent ascent (GDA) and stochastic gradient descent ascent (SGDA).
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Hashing Based Prediction for Large-Scale Kernel Machine–
Experimental results on most commonly used large-scale datasets show that the proposed algorithm outperforms the state-of-art kernel prediction methods in time cost while maintaining satisfactory accuracy.
- –
- –
- –
- Sensor assignment problem for vehicle unit cooperative tracking–
This paper studies the distribution of sensors and uses Compact Genetic Algorithm to solve the sensor assignment problem for vehicle unit and the simulation results show that the Compact genetic Algorithm has better optimization performance.
- –
- –
- The roller bearing fault diagnosis methods with harmonic wavelet packet and multi-classification relevance vector machine–
Experimental results show that the proposed feature extraction method is more effective than conventional method and the simplified OAO-RVM model possess the best fault discrimination accuracy and DT-R VM model possesses the better computation efficiency.
- –
- Roller bearing fault discrimination with harmonic wavelet package and ORO-RVM–
A novel roller bearing faults discrimination method with harmonic wavelet package and OAO-RVM (One Against One-Relevance Vector Machine) is addressed, which shows that the proposed feature extraction method is more effective than conventional method.
- –
- The Improvement Algorithm of Wavelet in Spectral Analysis–
This article is mainly about the application of wavelet in spectrum analysis and finds that there is frequency alias in sub-band reconstruction which is caused by Mallat algorithm.
- –
- Phrase Alignment Based on Combination of Multiple Strategies–
This paper proposes a new method to extract phrase translation pairs based on aligning Chinese words and English words in bilingual corpus with multiple alignment strategies, and experimental results indicate that the extracted phrasetranslation pairs achieve 63.07% at accuracy.
- –
- –
- Research of Abstracting Local Traffic Road Data in Google Maps–
This article deeply analyzes the traffic network features and path searching methods which Google Maps employed,amines the policy of abstracting local traffic data using those available objects and methods provided by Google Maps APIs, and designed the algorithms to get traffic data in some classical road.
- A Method of Wavelet Neural Network Construction Based on NARMAX Model–
This paper describes a method of wavelet neural network structure determination and the scaling coefficient estimate algorithm based on NARMAX model that can effectively raise the recognition accuracy and the recognition speed.
- –
- –
- –
- Design and Research on Mobile Data Service Platform Based on Unix Environment–
A solution of 3G Mobile Data Service Platform based on Unix environment is put forward, and this platform can be also applicated in the current conventional 2G and 2.5G mobile data network environment.
- A New Nonlinear Smoothing and Recognizing Method of Well-Log Time Series–
A simplified algorithm of smoothing well--log curves of oil-layers is presented, which can reduce the number of parameters needed to optimize and classify the shape of one peak of well-log curves into bell, case, egg, funnel and date core.
- A Low-Power Pipelined Implementation of 2D Discrete Wavelet Transform.–
This paper presents a low-power implementation of a 2-D biorthogonal DWT processor that uses residue number arithmetic and shows that the design is able to fit into a 1,000,000-gate FPGA device.
- –
- The Design and Implementation of Multimedia Courseware on Web–
The signification and implementation of Multimedia Courseware on Web is discussed, which will become the most popular courseware in future.
- –
- System of POL Curves Recognition Based on Wavelet Transformation–
A special function that can detect the singularity of signal by using wavelet transformation is introduced that was realized by computer pattern recognition technology and pattern recognition technique was introduced.
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.