Knowledge graph (KG) embedding is to embed components of a KG including entities and relations into continuous vector spaces, so as simplify the manipulation while preserving inherent structure KG. It can benefit variety downstream tasks such completion relation extraction, hence has quickly gained massive attention. In this article, we provide systematic review existing techniques, not only state-of-the-arts but also those with latest trends. Particularly, make based on type information...

In the last decades, due to development of parallel programming, lattice Boltzmann method (LBM) has attracted much attention as a fast alternative approach for solving partial differential equations. this paper, we first designed an energy functional based on fuzzy c-means objective function which incorporates bias field that accounts intensity inhomogeneity real-world image. Using gradient descent method, obtained corresponding level set equation from deduce external force LBM solver model...

In many applications, it is imperative to maintain high spectral and spatial resolution of remote sensing images. This letter addresses the issue by fusing low-spatial-resolution hyperspectral images (HSIs) high-spatial-resolution multispectral (MSIs) same scene collected coupled sensors and, thus, present a novel framework that generalizes well-established pan-sharpening algorithms. The main steps are dividing spectrum HSIs into several regions MSIs in each region chosen algorithm. Ratio...

In this paper, a fast pixel-level adapting background detection algorithm is presented. The proposed model records not only each pixel's historical values, but also estimates the efficacies of these based on occurrence statistics. It therefore capable removing least useful values from model, selectively to changes with different timescales, and restraining generation ghosts. A further control process adjusts individual decision threshold for pixel, reduces high frequency temporal noise,...

This paper presents a new image segmentation method that applies an edge-based level set in relay fashion. The proposed segments series of nested subregions are automatically created by shrinking the stabilized curves their previous subregions. final result is obtained combining all boundaries detected these has following three advantages: 1) It can be executed without human-computer interactions; 2) it with fashion to detect boundaries; and 3) obtains full specifying number relays advance....

This paper presents a new region-based unified tensor level set model for image segmentation. introduces three-order to comprehensively depict features of pixels, e.g., gray value and the local geometrical features, such as orientation gradient, then, by defining weighted distance, we generalized representative method from scalar tensor. The proposed has four main advantages compared with traditional follows. First, involving Gaussian filter bank, is robust against noise, particularly salt-...

In this paper, we present a novel level set method (LSM) for image segmentation. By utilizing the Bayesian rule, design nonlinear adaptive velocity and probability-weighted stopping force to implement robust segmentation objects with weak boundaries. The proposed is featured by following three properties: 1) it automatically determines curve shrink or expand rule involve regional features of images; 2) drives evolve an appropriate speed avoid leakage at boundaries; 3) reduces influence false...

<abstract> The survival rate of cervical cancer can be improved by the early screening. However, screening is a heavy task for pathologists. Thus, automatic cell classification model proposed to assist pathologists in In classification, number abnormal cells small, meanwhile, ratio between and normal small too. order deal with sample class imbalance problem, generative adversarial network (GAN) trained images obtain generated cells. Using both real images, convolutional neural (CNN) trained....

Deep learning (DL)-based modulation recognition methods are challenging in the case of few labeled samples and underwater impulsive noise. In this letter, we propose a novel network structure named IAFNet to achieve higher accuracy signals with fewer noise environment. The integrates preprocessing (INP), attention (AN) few-shot (FSL) extract features more effectively through denoising task-driven. Experimental results on simulation practical data show that attains stronger anti-noise...

The mainstream methods for change detection in synthetic-aperture radar (SAR) images use difference to define the initial regions. However, can suffer from semantic collapse, which makes it difficult determine information about changes. In this paper, we proposed a hierarchical fusion SAR image change-detection model based on conditional random field (HF-CRF). This introduces multimodal and constructs energy potential function using dynamic convolutional neural networks sliding window...

Recent years, multi-hop reasoning has been widely studied for knowledge graph (KG) due to its efficacy and interpretability. However, previous approaches are subject two primary shortcomings. First, agents struggle learn effective robust policies at the early phase sparse rewards. Second, these often falter on specific datasets like graphs, where required traverse lengthy paths. To address problems, we propose a model with dual based hierarchical reinforcement learning (HRL), which is named...

Abstract Deep Neural Networks (DNNs) have demonstrated outstanding performance in various medical image processing tasks. However, recent studies revealed a heightened vulnerability of DNNs to adversarial attacks compared their natural counterparts. In this work, we present novel perspective by analyzing the disparities between datasets and datasets, specifically focusing on dataset collection process. Our analysis uncovers unique differences data distribution across different classes...

This paper describes a master-slave teleoperation system which is developed to evaluate the effectiveness of teleopresence in telerobotics applications. The operator wears dataglove augmented with an arm-grounded force feedback device control dexterous hand and utilizes spaceball robot arm. Contact forces measured by finger sensors can be visual telepresence systems collect remote operation scenes display stereo helmet. A primitive autonomous grasp based on parallel joint torque/position...

Traditional spectral mixture analysis assumes that each endmember must have a constant signature. However, variability always exists in practical situations, which results reducing the accuracy of decomposition mixed pixels. In order to solve this problem, letter proposes new method based on Fisher discriminant null space (FDNS) for pixels hyperspectral imagery. The FDNS searches linear transformation spectra, makes those spectra no inside group but large differences among different groups....

In this paper, inclusion degree of fuzzy sets is introduced to image segmentation. The segmentation problem novelly modeled as the minimization overlapping rates between inside and outside regions, subject a constraint on total length region boundaries. Considering similar properties statistical domain, we use membership functions represent regions utilize nonparametric density estimates estimate them. Then, adopted formulate regions. We solve inclusion-degree-based optimization by deriving...

Federated learning has been recognized as a promising scheme to tackle the privacy issues in multi-access edge computing through periodically uploading machine (ML) model updates instead of original user data server. However, there still remains leakage such federated (FEL) systems since accessed by server can be utilized recover data. In this paper, we consider personalized differential based FEL alleviate adding different noise perturbations each device. Note that may degrade ML...

Due to its intrinsic nature which allows easily handle complex shapes and topological changes, the level set method (LSM) has been widely used in image segmentation. Nevertheless, LSM is computationally expensive, limits applications real-time systems. For this purpose, we propose a new algorithm, uses simultaneously edge, region, 2D histogram information order efficiently segment objects of interest given scene. The computational complexity proposed greatly reduced by using highly...

This paper presents a hybrid level set method for object segmentation. The deconstructs segmentation task into two procedures, i.e., shape transformation and curve evolution, which are alternately optimized until convergence. In this framework, only one prior encoded by context is utilized to estimate allowing the have same semantic expression as prior, evolution driven an energy functional with topology-preserving kernelized terms. such way, proposed featured following advantages: 1)...