Histopathology, particularly hematoxylin and eosin (H\&E) staining, plays a critical role in diagnosing characterizing pathological conditions by highlighting tissue morphology. However, H\&E-stained images inherently lack molecular information, requiring costly resource-intensive methods like spatial transcriptomics to map gene expression with resolution. To address these challenges, we introduce HECLIP (Histology-Enhanced Contrastive Learning for Imputation of Profiles), an innovative deep...

Data mining is one hot orientation in today's research field. Human activity recognition meaningful our daily living and a significant aspect data mining. Most previously almost based on tri-axial accelerometer. This paper presents novel method to collect from both accelerometer gyroscope using smartphone. Our activities including Walking, Running, Upstairs, Downstairs, Standing, Sitting Cycling, total of seven categories are classified. The raw MEMS recorded by smartphone according...

This paper examines the recent upsurge in foreign acquisitions of U.S. firms, specifically focusing on made by firms located emerging markets.Neoclassical theory predicts that, net, capital should flow from countries that are capital-abundant to capital-scarce.Yet increasingly market acquiring assets developed countries.Using transaction-specific acquisition data and firm-level accounting we evaluate post-acquisition performance publicly traded have been acquired markets over period...

In this paper, we present a new moving objects detection method in dynamic scenes to meet the real-time requirements. method, propose an improved SURF algorithm for feature extraction. The is via limiting number of detected points, and adopting fast reduce repeated calculation when calculating point's dominant orientation. This improves speed precision original SURF. To computation complexity global-search matching proposed. reduces time matching. Experimental results demonstrate that our...

The common method of image classification based on traditional SIFT local feature description makes the global information not comprehensive and has complicated calculation because construction scale extreme space. In addition, space is high dimensional sparse which will result in low accuracy, data redundancy time-consuming process. paper adopts a new with Bag-of-Words model improved algorithm. Each divided into lot uniform grid patches single descriptor 128 extracted each patch. Then...

Path planning is always an essential issue and complicated optimum problem for unmanned aerial vehicle (UAV). Genetic algorithms are well applied to solve such problems as a stochastic search method. In this paper, new method of path UAV based on genetic introduced. Reasonable coding way fitness function used in improved algorithm, prior knowledge added the algorithm. By selecting points moving strategy advance, can highly reduce computation cost find optimal more efficiently. The simulation...

In recent years, deep convolutional neural networks have made breakthrough progress in object recognition and detection tasks the field of computer vision, achieved great results both accuracy speed. However, small objects is still difficult detection, on common dataset MS COCO very low. This paper briefly reviews some work multi-scale algorithms, then proposes a method feature enhancement fusion based maps, improving COCO.

The service sector has become increasingly important in today's economy. To meet the rising expectation of high‐quality services, efficiently allocating resources is vital for systems to balance qualities with costs. In particular, this paper focuses on a class resource allocation problems where service‐level objective and constraints are form probabilistic measures. Further, process complexity system dynamics often render their performance evaluation optimization challenging relying...

When Multi-DSP parallel architecture transfers to distributed memory way from shared way, its parallelism with fine-grained become weak, and it's difficult offer SIFT's complex computing satisfy the need of real-time. In paper, a algorithm, based on homogeneous Multi-core DSP, referring DSM model processing machines is presented. Firstly, master processor separates task into several small tasks by exploiting coarse-grained inherent; then, through high-speed network for data-exchange, each...

This paper aims to explore how the students’ leadership skills would be developed through team work activities in classes. As definition of is diverse and elusive, it hard evaluate whether students have been improved getting involved work. So, by extracting core commonly recognized both academic business world, questionnaires interviews are designed made among two groups students: teacher-students vocational college students. Based on data gathered, it’s agreed that can cultivated...

We investigate a deadline-constrained task scheduling problem in the fog computing environments where tasks can be offloaded to heterogeneous resources. Three kinds of resources are involved: mobile device, device and cloud server. The objective is schedule all with minimum energy consumption. develop an energy-aware strategy propose critical path based iterative algorithm which obtain optimal solution polynomial time complexity. also discuss cases when no feasible exists. Experimental...

Mobile ad hoc network (MANET) has attracted a lot of research attention. In the analysis mobile simulation, choice node mobility model is very important. military battlefield, group occurs frequently and nodes in usually locate an elliptic region. Therefore, we propose new model-Ellipse Group Mobility (EGM) model. And compared performance EGM with Reference Point (RPGM) Simulation results show that more suitable for tactical MANET.

This paper examines the recent upsurge in foreign acquisitions of U.S. firms, specifically focusing on made by firms located emerging markets. Neoclassical theory predicts that, net, capital should flow from countries that are capital-abundant to capital-scarce. Yet increasingly market acquiring assets developed countries. Using transaction-specific acquisition data and firm-level accounting we evaluate post-acquisition performance publicly traded have been acquired markets over period...

TV logo recognition plays an important role in video content understanding. So far, Real-time under complex background based on single frame is still a very challenge task. By analyzing the characteristics of logo, new algorithm multiple feature fusion via hierarchical matching was proposed this paper. The process consisted three stages. Firstly, coarse differentiation and normalized cross correlation coefficient employed to narrow down candidate space significantly; then fine HOG used...

Foreground detection in dynamic background has become a hot topic video surveillance recent years. this paper we propose new foreground approach based on GPU background. with the proposed method, SIFT features are first extracted from two adjacent frames sequences, which can be utilized to compute parameters of affine transform model and solve global motion compensation. then improving subtraction updating module is adopted detect objects. method used improve application performance....

Confronted with the task environment full of repetitive textures, state-of-art description and detection methods for local features greatly suffer from "pseudo-negatives," bringing inconsistent optimization objectives during training. To address this problem, article develops a self-supervised graph-based contrastive learning framework to train model features, GCLFeat. The proposed approach learns alleviate pseudo-negatives specifically three aspects: 1) designing graph neural network (GNN),...

Object detection algorithm based on depth model has achieved state-of-the-art results various challenging benchmarks. However, the large amount of parameters means a calculation. This seriously limits practical application object algorithm, especially embedded devices with limited computing power. We propose an DenseNet and Region Proposal Network(RPN) replace ROI Pooling Align. From evaluation PASCAL VOC MS COCO we can see that achieves fewer while maintaining or improving accuracy. is...

An algorithm namely the TDOA-Camberra is proposed for multi-target passive location in this paper. The uses Camberra distance to associate target data and do optimal search, then time difference of arrival (TDOA) locate multiple targets passively once from same determined. At time, as computing associating using long, improved sequential similarity detection (SSDA) added calculation order shorten association, thus increase speed location. theoretical analysis simulation experiment proves...

In this paper we present a fast vision-based eye-gaze tracking method based on Particle filtering algorithm in the condition of near-infrared light and single-camera, against to requirement real-time eye engineering, fact that presently most methods video are not precise, target easy lose. initialize step, use high accuracy cascaded classifier trained by AdaBoost get primitive information region. Considering region last frame image is valuable next analysis, particle filter adopted...

Black-box problems are common in real life like structural design, drug experiments, and machine learning. When optimizing black-box systems, decision-makers always consider multiple performances give the final decision by comprehensive evaluations. Motivated such practical needs, we focus on constrained where objective constraints lack known special structure, evaluations expensive even with noise. We develop a novel Bayesian optimization approach based knowledge gradient method...