The results of chest X-ray (CXR) analysis 2D images to get the statistically reliable predictions (availability tuberculosis) by computer-aided diagnosis (CADx) on basis deep learning are presented. They demonstrate efficiency lung segmentation, lossless and lossy data augmentation for CADx tuberculosis convolutional neural network (CNN) applied small not well-balanced dataset even. CNN demonstrates ability train (despite overfitting) pre-processed obtained after segmentation in contrast...

Uterine cancer, also known as endometrial can seriously affect the female reproductive organs, and histopathological image analysis is gold standard for diagnosing cancer. However, due to limited capability of modeling complicated relationships between images their interpretations, these computer-aided diagnosis (CADx) approaches based on traditional machine learning algorithms often failed achieve satisfying results. In this study, we developed a CADx approach using convolutional neural...

The efficiency of some dimensionality reduction techniques, like lung segmentation, bone shadow exclusion, and t-distributed stochastic neighbor embedding (t-SNE) for exclusion outliers, is estimated analysis chest X-ray (CXR) 2D images by deep learning approach to help radiologists identify marks cancer in CXR. Training validation the convolutional neural network (CNN) was performed on open JSRT dataset (dataset #01), after - BSE-JSRT #02), segmentation #03), #04), segmented outliers t-SNE...

By applying the time-delay control theory to a TCP/RED dynamic model, this note establishes some explicit conditions under which system is stable in terms of average queue length. Then, stability region discussed. Finally, results are illustrated by using ns2 simulations, demonstrates that it able choose an appropriate parameter max/sub p/ RED based on derived note, achieve satisfactory network performance. It found, comparison, improved performance better than three other typical active...

Adaptive Monte Carlo localization (AMCL) algorithm has a limited pose accuracy because of the nonconvexity laser sensor model, complex and unstructured features working environment, randomness particle sampling, final selection problem. In this paper, an improved AMCL is proposed, aiming to build radar‐based robot system in with LIDAR point cloud scan‐matching process after score calculating process. The weighted mean swarm used as initial scan matching matched probability grid map from...

Robot intelligent inspection is widely used in the positioning of various pointer instruments power, petroleum, chemical, and other industries. Aiming at technical problems poor adaptability, real-time performance, low accuracy instrument method existing substation robot system, we propose a simple effective detection algorithm. The algorithm first extracts locally adaptive regression kernels (LARK) features input image, dimension LARK feature reduced using principal components analysis...

ABSTRACT In recent years, the rapid advancement of automation control and intelligent sensing technologies has positioned autonomous driving as a focal point interest for both academia industry. As core equipment in modern construction industrial production, engineering machinery urgently requires transformation. To promote development machinery, we have designed an integrated bulldozer system, which can be extended to various types machinery. For specific mine dumping operation environment,...

For the existing visual-inertial SLAM algorithm, when robot is moving at a constant speed or purely rotating and encounters scenes with insufficient visual features, problems of low accuracy poor robustness arise. Aiming to solve inertial tightly coupled vision-IMU-2D lidar odometry (VILO) algorithm proposed. Firstly, low-cost 2D observations are fused in manner. Secondly, model used derive Jacobian matrix residual respect state variable be estimated, constraint equation constructed....

To realize a robust robotic grasping system for unknown objects in an unstructured environment, large amounts of grasp data and 3D model the object are required; sizes these directly affect rate successful grasps. reduce time cost acquisition labeling increase grasps, we developed self-supervised learning mechanism to control tasks performed by manipulators. First, manipulator automatically collects point cloud from multiple perspectives efficiency acquisition. The complete is obtained using...

A method based on the one-dimensional local binary pattern (1-D LBP) algorithm to extract features of ultrasonic defect signals and perform multi-class classification was proposed. The echo were first decomposed into wavelet coefficients by packet decomposition. 1-D LBP employed components at low high frequencies, respectively. Subsequently, these statistical feature sets regarded as vectors classification. Weld defects then classified automatically using radial basis function support vector...

The dissolved gas in transformer oil, which could represents the faults, can be analyzed by spectroscopy. Since BP neural network method involves a large amount of data matrices operation leads to much computation and that single computer not meet requirements real-time analysis. In order improve situation, this paper proposes quantitative spectral analysis oil based on parallel network. This designs model builds independently Hadoop clustering computing platform implement model. cluster...

Prioritized experience replay (PER) chooses the data based on value of Temporal-Difference (TD) error, it can improve utilization in deep reinforcement learning methods. But since TD error needs to be calculated when sampled at each time, so computational complexity PER is high. And hyperparameters also affect process. The need adjusted carefully. To cope above problems, we propose meta-learning-based buffer separation (MSER) this paper. Firstly, original divided into a successful and...

In the field of intelligent manufacturing, robot grasping and sorting is important content. However, there are some disadvantages in traditional single-view-based manipulator methods by using a 2D camera, where efficiency accuracy both low when facing scene stacking occlusion for reason that information missing single-view camera-based while acquiring information, only can't change difficult-to-grasp which stack occluded. Regarding issue above, pushing-grasping collaborative method based on...

Normal operation of the transformer is an important guarantee reliability power system. The fault diagnosis basis for maintenance. Three-ratio method widely used in oil-immersed diagnosis, however encoded value too severe to correspond type failure. FCM (Fuzzy C-means clustering) algorithm introduced solve this problem, and its performance determines correct rate diagnosis. This paper focuses on defects view diagnostic data characteristic three-ratio method, optimized two ways. Firstly,...

The current Internet is dramatically suffering the Distributed Denial of Service (DDoS) attacks, in which perpetrator maliciously makes network resource unavailable to its intended users by temporarily or indefinitely disrupting services a host connected Internet. In this paper, we investigate an transmission control protocol/active queue management (TCP/AQM) router subject DDoS attacks. We utilize time delay theory analyze dynamics congestion windows, and queues at router. derive some...

(This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version no longer accessible.) To improve efficiency of deep reinforcement learning (DRL)-based methods robot manipulator trajectory planning in random working environments, we present three dense reward functions. These rewards differ from traditional sparse reward. First, a posture function is proposed speed up process with more reasonable by modeling distance...

Considering the adverse impact of speed measurement on accuracy pose estimation after a mobile robot slips, collides, or abducts, this paper proposes monocular inertial simultaneous localization and mapping algorithm that includes wheel anomaly detection. The adds to least squares problem in tightly coupled manner uses nonlinear optimization method maximize posterior probability solve optimal state estimation. For control Mecanum wheel, because existing closed-loop cannot calculate motion...