Intrusion detection systems (IDS) play an important role in the protection of network operations and services. In this paper, we propose effective intrusion scheme based on deep learning techniques. The proposed employs a denoising autoencoder (DAE) with weighted loss function for feature selection, which determines limited number features to reduce dimensionality. selected data is then classified by compact multilayer perceptron (MLP) identification. Extensive experiments are conducted...

Aiming at a class of high-order strict feedback nonlinear multi-agent systems with communication constraints, novel distributed adaptive back-stepping control method is proposed to cooperatively track the moving targets. First, five agents are used as controlled objects, and all form "leader-follower" mode structure. Meanwhile, leader's velocity considered forward whole formation system, remaining follow movement. Then, each agent tracks desired time-varying reference trajectory, thereby...

Tiny object detection (TOD) remains a challenging problem due to the extremely small size and weak feature presentations of tiny objects. Many effective methods have improved objects below <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$32\times 32$ </tex-math></inline-formula> pixels some extent, but performance is still poor for notation="LaTeX">$16\times 16$ pixels. In this paper, we find that aliasing...

Recently, there has been increasing interest in using deep learning techniques for various seismic interpretation tasks. However, unlike shallow machine models, models are often far more complex and can have hundreds of millions free parameters. This not only means that large amounts computational resources needed to train these but critically, they require vast labeled training data as well. In this work, we show how automatically-generated weak labels be effectively used overcome problem...

This article investigates the cooperative fault-tolerant tracking problem for multiagent systems with directed topology that experience communication link faults and actuator under malicious attacks. Based on a resilient event-triggered mechanism, novel distributed high-order sliding-mode observer robust adaptive control method are proposed, which can achieve desired formation shape limitations hybrid caused by First, designed estimate unknown leader. Second, proposed mechanism be used to...

Aerial images often depict objects with arbitrary orientations, which pose challenges for conventional object detectors to detect and classify. To address this issue, rotation-equivariant Convolutional Neural Networks (CNNs) have been proposed extract features. However, the orientation encoding in these networks is unstable noisy, deteriorating detection performance. In paper, we first analyze network. Then, propose a Rotation-robust Prototype Generation (RPG) method, consists of two parts,...

The timely detection of disease outbreaks through reliable early warning signals (EWSs) is indispensable for effective public health mitigation strategies. Nevertheless, the intricate dynamics real-world spread, often influenced by diverse sources noise and limited data in stages outbreaks, pose a significant challenge developing EWSs, as performance existing indicators varies with extrinsic intrinsic noises. Here, we address modelling when measurements are corrupted additive white noise,...

Event cameras are bio-inspired vision sensors measuring brightness changes (referred to as an 'event') for each pixel independently, instead of capturing images at a fixed rate using conventional cameras. Asynchronous event data mixed with noise information is challenging event-based tasks. In this paper, we propose broad learning network object detection the data. The consists two distinct layers, feature-node layer and enhancement-node layer. Different convolutional neural networks, can be...

Multipedestrian tracking in traffic scenes is challenging due to cluttered backgrounds and serious occlusions. In this paper, we propose a layered graph model image (RGB) depth (D) domains for real-time robust multipedestrian tracking. The motivation investigate high-level constraints RGB-D data association improve the optimization from trajectory level layer level. To construct graph, define domain so that pedestrian objects are assigned proper layers. We use detection responses RGB as...

In order to study the influence of pressure-equalizing exhaust at shoulder a submarine-launched vehicle on surface hydrodynamic characteristics, this paper establishes numerical calculation method based VOF multiphase flow model, standard RNG turbulence model and overset mesh technology; compares fusion characteristics air film underwater vehicle, as well distribution pressure along vehicle’s axial direction. The results show that approximate isobaric zone derived from can greatly improve...

Epidermal electrophysiological monitoring has garnered significant attention for its potential in medical diagnosis and healthcare, particularly continuous signal recording. However, simultaneously satisfying skin compliance, mechanical properties, environmental adaptation, biocompatibility to avoid attenuation motion artifacts is challenging, accurate physiological feature extraction necessitates effective signal-processing algorithms. This review presents the latest advancements smart...

Tracking multiple persons is a challenging task when move in groups and occlude each other. Existing group-based methods have extensively investigated how to make group division more accurately tracking-by-detection framework; however, few of them quantify the dynamics from perspective targets' spatial topology or consider dynamic view. Inspired by sociological properties pedestrians, we propose novel socio-topology model with topology-energy function factor moving groups. In this model,...

In engineering design optimization, the optimal sampling method is usually used to solve large-scale and complex system problems. A (FOLHD) of fast Latin hypercube proposed in order overcome time-consuming poor efficiency traditional methods. FOLHD algorithm based on inspiration that a near can be established by small-scale initial sample generated using Successive Local Enumeration Translational Propagation algorithm. Moreover, resizing strategy presented generate samples with arbitrary...

Weakly supervised object classification and localization are learned classes locations using only image-level labels, as opposed to bounding box annotations. Conventional deep convolutional neural network (CNN)-based methods activate the most discriminate part of an in feature maps then attempt expand activation whole object, which leads deteriorating performance. In addition, those use semantic information last map, while ignoring role shallow features. So, it remains a challenge enhance...

Unsupervised learning is a challenging task due to the lack of labels. Multiple Object Tracking (MOT), which inevitably suffers from mutual object interference, occlusion, etc., even more difficult without label supervision. In this paper, we explore latent consistency sample features across video frames and propose an Contrastive Similarity Learning method, named UCSL, including three contrast modules: self-contrast, cross-contrast, ambiguity contrast. Specifically, i) self-contrast uses...

Drone is widely applied to human production life with the development of drone technology. Therefore, it very advantageous carry out research on target detection in UAV images. Using technology, achieve accurate capture target, thereby achieving positioning and goal meeting various mission requirements. Unmanned aerial vehicle (UAV) images have characteristics large changes scale, small overall size, are limited by viewing angle, which pose a great challenge targets. This thesis mainly...