With the rapid development of blockchain technology, P2P networks are facing increasing security threats, among which Eclipse attacks, as a type of network isolation attack, have seriously affected the normal operatio...
详细信息
Gradient compression is a promising approach to alleviating the communication bottleneck in data parallel deep neural network (DNN) training by significantly reducing the data volume of gradients for synchronization. ...
详细信息
Gradient compression is a promising approach to alleviating the communication bottleneck in data parallel deep neural network (DNN) training by significantly reducing the data volume of gradients for synchronization. While gradient compression is being actively adopted by the industry (e.g., Facebook and AWS), our study reveals that there are two critical but often overlooked challenges: 1) inefficient coordination between compression and communication during gradient synchronization incurs substantial overheads, and 2) developing, optimizing, and integrating gradient compression algorithms into DNN systems imposes heavy burdens on DNN practitioners, and ad-hoc compression implementations often yield surprisingly poor system performance. In this paper, we propose a compression-aware gradient synchronization architecture, CaSync, which relies on flexible composition of basic computing and communication primitives. It is general and compatible with any gradient compression algorithms and gradient synchronization strategies and enables high-performance computation-communication pipelining. We further introduce a gradient compression toolkit, CompLL, to enable efficient development and automated integration of on-GPU compression algorithms into DNN systems with little programming burden. Lastly, we build a compression-aware DNN training framework HiPress with CaSync and CompLL. HiPress is open-sourced and runs on mainstream DNN systems such as MXNet, TensorFlow, and PyTorch. Evaluation via a 16-node cluster with 128 NVIDIA V100 GPUs and a 100 Gbps network shows that HiPress improves the training speed over current compression-enabled systems (e.g., BytePS-onebit, Ring-DGC and PyTorch-PowerSGD) by 9.8%-69.5% across six popular DNN models. IEEE
A lot of research shows that there could be several reasons why the duality of agricultural products has been reduced. Plant diseases make up one of the most important components of this quality. Therefore, the reduct...
详细信息
Cyberbullying presents a profound and pressing challenge in our society today, particularly in the context of the burgeoning era of Web 4.0. Despite the remarkable advancements in technology and software applications,...
详细信息
Creating natural language descriptions or captions for images is a formidable task that requires a combination of computer vision techniques to understand image content and natural language processing models to expres...
详细信息
Over the past decades, integration of wireless sensor networks (WSNs) and computer vision (CV) technology has shown promising results in mitigating crop losses caused by wild animal attacks. Studies have demonstrated ...
详细信息
Over the past decades, integration of wireless sensor networks (WSNs) and computer vision (CV) technology has shown promising results in mitigating crop losses caused by wild animal attacks. Studies have demonstrated the effectiveness of these technologies in providing real-time monitoring and early detection of animal intrusions into agricultural fields. By deploying WSNs equipped with motion sensors and cameras, farmers can receive instant alerts when wild animals enter their fields, allowing for timely intervention to prevent crop damage. Furthermore, advancements in CV algorithms possess made possible to automatically detect and classify the animal species, facilitating targeted response strategies. For example, sophisticated image processing techniques can differentiate between harmless birds and destructive mammals, allowing farmers to focus their efforts on deterring the most damaging species. Field trials and pilot projects implementing WSN-CV systems have reported significant reductions in crop losses attributed to wild animal raids. By leveraging data collected through sensor networks and analyzed using computer vision algorithms, farmers can make informed decisions regarding pest and insect management strategies. This data-driven approach has led to more efficient utilization of resources, such as targeted application of insecticides and pesticides, resulting in both economic and environmental benefits. Moreover, the integration of WSN-CV technology has enabled the development of innovative deterrent systems that leverage artificial intelligence and automation. These systems can deploy non-lethal methods, such as sound or light-based repellents, to deter wild animals without causing harm to the environment or wildlife populations. Overall, the combination of wireless sensor networks and computer vision technology provides the promising resolution to the long-standing issue of wild animal-related losses in agriculture. By harnessing the power of data and a
The Internet of Medical Things in healthcare necessitates a secure way to transfer medical images due to the rise of telemedicine. Security breaches in public networks lead to falsification of medical images and furth...
详细信息
This paper presents a remotely operated robotic system that includes two mobile manipulators to extend the functional capabilities of a human *** with previous tele-operation or robotic body extension systems,using tw...
详细信息
This paper presents a remotely operated robotic system that includes two mobile manipulators to extend the functional capabilities of a human *** with previous tele-operation or robotic body extension systems,using two mobile manipulators helps with enlarging the workspace and allowing manipulation of large or long *** system comprises a joystick for controlling the mobile base and robotic gripper,and a motion capture system for controlling the arm *** together enable tele-operated dual-arm and large-space *** the experiments,a human tele-operator controls the two mobile robots to perform tasks such as handover,long object manipulation,and cooperative *** results demonstrated the effectiveness of the proposed system,resulting in extending the human body to a large space while keeping the benefits of having two limbs.
Colon cancer is the third most commonly diagnosed cancer in the *** colon AdenoCArcinoma(ACA)arises from pre-existing benign polyps in the mucosa of the ***,detecting benign at the earliest helps reduce the mortality ...
详细信息
Colon cancer is the third most commonly diagnosed cancer in the *** colon AdenoCArcinoma(ACA)arises from pre-existing benign polyps in the mucosa of the ***,detecting benign at the earliest helps reduce the mortality *** this work,a Predictive Modeling System(PMS)is developed for the classification of colon cancer using the Horizontal Voting Ensemble(HVE)*** different patterns inmicroscopic images is essential to an effective classification system.A twelve-layer deep learning architecture has been developed to extract these *** developedHVE algorithm can increase the system’s performance according to the combined models from the last epochs of the proposed *** thousand(10000)microscopic images are taken to test the classification performance of the proposed PMS with the HVE *** microscopic images obtained from the colon tissues are classified intoACAor benign by the proposed *** prove that the proposed PMS has∼8%performance improvement over the architecture without using the HVE *** proposed PMS for colon cancer reduces the misclassification rate and attains 99.2%of sensitivity and 99.4%of *** overall accuracy of the proposed PMS is 99.3%,and without using the HVE method,it is only 91.3%.
暂无评论