检索结果-内蒙古大学图书馆

Reinforcement Learning Scheduling Research for Edge Servers

学校读者我要写书评

暂无评论

Reinforcement Learning Scheduling Research for Edge Servers

International Conference on computer Network, Electronic and Automation (ICCNEA)

作者： Yonggui Han Zhongsheng Wang School of Computer Science and Engineering Xi’an Technological University Xi’an China State and Provincial Joint Engineering Lab. of Advanced Network Monitoring and Control Xi'an Technological University Xi'an China

The rapid development of the Internet of Things and big data has brought about massive amounts of data and the need for real-time responses, which traditional computing models can no longer meet. Edge computing comes into being, which pushes computing resources and services closer to the source of data generation so that data can be processed and analyzed closer to the user. The scheduling of siting strategies for edge servers is crucial for improving the performance of edge computing networks. Traditional scheduling methods for siting strategies are often based on rules or optimization algorithms, which lack adaptive and intelligent management of the network *** paper proposes a reinforcement learning DQN-based edge server siting policy scheduling method, which aims to achieve intelligent management and resource optimization of edge computing networks by learning the optimal policy. Experimental results show that the method has significant performance advantages in terms of delay, load balancing and energy consumption. The experiment adopts DQN algorithm, which is based on Q-learning and guides the agent to make decisions by optimizing Q-value function. In this paper, the reinforcement learning scheduling algorithm of edge server is studied and compared with other three algorithms. According to the experimental results, the proposed reinforcement learning scheduling algorithm reduces the delay by 7%, the energy consumption by 3%, and the load balancing by 20% compared with the other three algorithms. This shows that the algorithm has significant advantages in improving system performance.

关键词：

BRNES: Enabling Security and Privacy-aware Experience Sharing in Multiagent Robotic and Autonomous Systems

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Hossain, Md Tamjid La, Hung Manh Badsha, Shahriar Netchaev, Anton The Advanced Robotics and Automation Lab Computer Science and Engineering University of Nevada RenoNV United States Bosch Engineering North America The U.S. Army Corps United States

Although experience sharing (ES) accelerates multiagent reinforcement learning (MARL) in an advisor-advisee framework, attempts to apply ES to decentralized multiagent systems have so far relied on trusted environments and overlooked the possibility of adversarial manipulation and inference. Nevertheless, in a real-world setting, some Byzantine attackers, disguised as advisors, may provide false advice to the advisee and catastrophically degrade the overall learning performance. Also, an inference attacker, disguised as an advisee, may conduct several queries to infer the advisors’ private information and make the entire ES process questionable in terms of privacy leakage. To address and tackle these issues, we propose a novel MARL framework (BRNES) that heuristically selects a dynamic neighbor zone for each advisee at each learning step and adopts a weighted experience aggregation technique to reduce Byzantine attack impact. Furthermore, to keep the agent’s private information safe from adversarial inference attacks, we leverage the local differential privacy (LDP)-induced noise during the ES process. Our experiments show that our framework outperforms the state-of-the-art in terms of the steps to goal, obtained reward, and time to goal metrics. Particularly, our evaluation shows that the proposed framework is 8.32x faster than the current non-private frameworks and 1.41x faster than the private frameworks in an adversarial setting. © 2023, CC BY-NC-SA.

关键词： Multi agent systems

Research and Implementation of Image Defogging Based on Deep Learning

学校读者我要写书评

暂无评论

Research and Implementation of Image Defogging Based on Deep...

International Conference on computer Network, Electronic and Automation (ICCNEA)

作者： Ao Li Zhongsheng Wang School of Computer Science and Engineering Xi’an Technological University Xi’an China State and Provincial Joint Engineering Lab. of Advanced Network Monitoring and Control Xi'an Technological University Xi'an China

Image degradation due to fog and haze is a common problem in the field of life and image research. This article will be designed and implemented of image haze removal algorithm. The traditional methods have some problems such as low brightness, artifact, halo and distortion, but the existing methods based on deep learning have better effect. In this paper, we will propose a de-fogging network model, which is based on adversarial network. The generator in adversarial network is a convolutional neural network with residual network, which reduces the model training parameters and improves the parameter utilization. The discriminator uses the full convolutional network as the discriminator, which enhances the spatial sensitivity of the model to the image. In order to improve efficiency, the discriminator is simplified on the original structure. The loss function of the model is improved so that the model can get better training effect.

关键词：

Research on Multilevel Degradater Model under Unsupervised Learning

学校读者我要写书评

暂无评论

Research on Multilevel Degradater Model under Unsupervised L...

International Conference on computer Network, Electronic and Automation (ICCNEA)

作者： Jieyi Lv Zhongsheng Wang School of Computer Science and Engineering Xi'an Technological University Xi'an China State and Provincial Joint Engineering Lab. of Advanced Network Monitoring and Control Xi'an Technological University Xi'an China

In most super-resolution experiments, because the data set in the real scene cannot meet the quantity requirements, it is necessary to downsample the high-resolution images using the degradation method to obtain the same proportion of low-resolution images. However, because the same degradation method is used to process images, the processing process of low-resolution images is relatively simple, which is easy to cause model solidification in the subsequent training process. After the training is completed, the generalization ability will be generally low due to the single underlying degradation method. This paper uses the SRMD method to build an unsupervised multilevel degradation, and takes advantage of the adduction network generated in unsupervised learning. After extracting image features, it adopts anisotropic Gaussian fuzzy kernel, bi-cubic subsampling and additive Gaussian noise to compose the degradation training data set. Compared with the traditional bi-cubic PNSR, the performance is improved by 5%. The SSIM value is increased by 3%, and the running speed is increased by 7% on average.

关键词：

Fine-grained Background Representation for Weakly Supervised Semantic Segmentation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Yin, Xu Im, Woobin Min, Dongbo Huo, Yuchi Pan, Fei Yoon, Sung-Eui The School of Computing Korea Advanced Institute of Science and Technology Daejeon34141 Korea Republic of The Faculty of the Department of Computer Science and Engineering Ewha Womans University Seoul03760 Korea Republic of The State Key Lab of CAD and CG Zhejiang University China Zhejiang Lab 310058 China The School of Computer Science Engineering University of Michigan United States The Faculty of School of Computing Korea Advanced Institute of Science and Technology Deajeon34141 Korea Republic of

Generating reliable pseudo masks from image-level labels is challenging in the weakly supervised semantic segmentation (WSSS) task due to the lack of spatial information. Prevalent class activation map (CAM)-based solutions are challenged to discriminate the foreground (FG) objects from the suspicious background (BG) pixels (a.k.a. co-occurring) and learn the integral object regions. This paper proposes a simple fine-grained background representation (FBR) method to discover and represent diverse BG semantics and address the co-occurring problems. We abandon using the class prototype or pixel-level features for BG representation. Instead, we develop a novel primitive, negative region of interest (NROI), to capture the fine-grained BG semantic information and conduct the pixel-to-NROI contrast to distinguish the confusing BG pixels. We also present an active sampling strategy to mine the FG negatives on-the-fly, enabling efficient pixel-to-pixel intra-foreground contrastive learning to activate the entire object region. Thanks to the simplicity of design and convenience in use, our proposed method can be seamlessly plugged into various models, yielding new state-of-the-art results under various WSSS settings across benchmarks. Leveraging solely image-level (I) labels as supervision, our method achieves 73.2 mIoU and 45.6 mIoU segmentation results on Pascal Voc and MS COCO test sets, respectively. Furthermore, by incorporating saliency maps as an additional supervision signal (I+S), we attain 74.9 mIoU on Pascal Voc test set. Concurrently, our FBR approach demonstrates meaningful performance gains in weakly-supervised instance segmentation (WSIS) tasks, showcasing its robustness and strong generalization capabilities across diverse domains. Copyright © 2024, The Authors. All rights reserved.

关键词： Semantic Segmentation

SOFT INJECTABLES USING SMART THREADS FOR DOSE-CONTROLLED DRUG DELIVERY

学校读者我要写书评

暂无评论

SOFT INJECTABLES USING SMART THREADS FOR DOSE-CONTROLLED DRU...

2022 Solid-State Sensors, Actuators and Microsystems Workshop, Hilton Head 2022

作者： Alsaedi, Mossab K. Das, Riddha Nejad, Hojatollah Rezaei Sharma, Atul Sonkusale, Sameer Nano Lab. Advanced Technology Laboratory Tufts University MedfordMA United States Department of Chemical and Biological Engineering Tufts University MedfordMA United States Department of Electrical and Computer Engineering Tufts University MedfordMA United States Department of Biomedical Engineering Tufts University MedfordMA United States

ISBN: (纸本)9781940470047

Transdermal drug delivery is a promising alternative to conventional drug delivery approaches, such as oral ingestion and hypodermic injection. Hypodermic injections are often painful, and oral ingestion of drugs requires the administration of doses higher than the needed to compensate for enzymatic degradation and poor absorption of drugs. Current transdermal drug-delivery platforms include microneedles, which resolve the pain aspect of conventional drug delivery approaches. However, microneedle patches are impractical as they are easily peeled off with motion and sweat. In this study, we propose soft injectables comprised of sutures that are loaded with drugs for painless and localized sustained drug delivery in the dermis layer of the skin. They stay in place during the course of delivery and are suitable for any skin type. © 2022 TRF.

关键词： Controlled drug delivery

Integrating Prior Knowledge with Graph Encoder for Gene Regulatory Inference from Single-cell RNA-Seq Data

学校读者我要写书评

暂无评论

Integrating Prior Knowledge with Graph Encoder for Gene Regu...

2022 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2022

作者： Li, Jiawei Yang, Fan Wang, Fang Rong, Yu Zhao, Peilin Chen, Shizhan Yao, Jianhua Tang, Jijun Guo, Fei Tianjin University College of Intelligence and Computing Tianjin China AI Lab Tencent Shenzhen China Shenzhen Institute of Advanced Technology Chinese Academy of Sciences Shenzhen China Central South University School of Computer Science and Engineering Changsha China

ISBN: (纸本)9781665468190

Inferring gene regulatory networks based on single-cell transcriptomes is critical for systematically understanding cell-specific regulatory networks and discovering drug targets in tumor cells. Here we show that existing methods mainly perform co-expression analysis and apply the image-based model to deal with the non-euclidean scRNA-seq data, which may not reasonably handle the dropout problem and not fully take advantage of the validated gene regulatory topology. We propose a graph-based end-to-end deep learning model for GRN inference (GRNInfer) with the help of known regulatory relations through transductive learning. The robustness and superiority of the model are demonstrated by comparative experiments. © 2022 IEEE.

关键词： RNA

Automatic Scene Generation: State-of-the-Art Techniques, Models, Datasets, Challenges, and Future Prospects

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Fime, Awal Ahmed Mahmud, Saifuddin Das, Arpita Islam, Md Sunzidul Kim, Hong-Hoon Advanced Telerobotics Research Lab Kent State University KentOH44240 United States Bradley University PeoriaIL61625 United States Department of Computer Science and Engineering Khulna University of Engineering & Technology Khulna9203 Bangladesh

Automatic scene generation is an essential area of research with applications in robotics, recreation, visual representation, training and simulation, education, and more. This survey provides a comprehensive review of the current state-of-the-arts in automatic scene generation, focusing on techniques that leverage machine learning, deep learning, embedded systems, and natural language processing (NLP). We categorize the models into four main types: Variational Autoencoders (VAEs), Generative Adversarial Networks (GANs), Transformers, and Diffusion Models. Each category is explored in detail, discussing various sub-models and their contributions to the field. We also review the most commonly used datasets, such as COCO-Stuff, Visual Genome, and MS-COCO, which are critical for training and evaluating these models. Methodologies for scene generation are examined, including image-to-3D conversion, text-to-3D generation, UI/layout design, graph-based methods, and interactive scene generation. Evaluation metrics such as Fréchet Inception Distance (FID), Kullback-Leibler (KL) Divergence, Inception Score (IS), Intersection over Union (IoU), and Mean Average Precision (mAP) are discussed in the context of their use in assessing model performance. The survey identifies key challenges and limitations in the field, such as maintaining realism, handling complex scenes with multiple objects, and ensuring consistency in object relationships and spatial arrangements. By summarizing recent advances and pinpointing areas for improvement, this survey aims to provide a valuable resource for researchers and practitioners working on automatic scene generation. © 2024, CC BY.

关键词： Generative adversarial networks

Correcting Noisy Multilabel Predictions: Modeling label Noise through Latent Space Shifts

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Huang, Weipeng Li, Qin Xiao, Yang Qiao, Cheng Cai, Tie Liang, Junwei Hurley, Neil J. Piao, Guangyuan School of Computer Science and Software Engineering Shenzhen Institute of Information Technology Guangdong China State Key Lab of Integrated Service Networks School of Cyber Engineering Xidian University Shaanxi China Cyberspace Institute of Advanced Technology Guangzhou University Guangdong China School of Computer Science University College Dublin Dublin Ireland

Noise in data appears to be inevitable in most realworld machine learning applications and would cause severe overfitting problems. Not only can data features contain noise, but labels are also prone to be noisy due to human input. In this paper, rather than noisy label learning in multiclass classifications, we instead focus on the less explored area of noisy label learning for multilabel classifications. Specifically, we investigate the post-correction of predictions generated from classifiers learned with noisy labels. The reasons are two-fold. Firstly, this approach can directly work with the trained models to save computational resources. Secondly, it could be applied on top of other noisy label correction techniques to achieve further improvements. To handle this problem, we appeal to deep generative approaches that are possible for uncertainty estimation. Our model posits that label noise arises from a stochastic shift in the latent variable, providing a more robust and beneficial means for noisy learning. We develop both unsupervised and semi-supervised learning methods for our model. The extensive empirical study presents solid evidence to that our approach is able to consistently improve the independent models and performs better than a number of existing methods across various noisy label settings. Moreover, a comprehensive empirical analysis of the proposed method is carried out to validate its robustness, including sensitivity analysis and an ablation study, among other elements. © 2025, CC BY.

关键词： Contrastive Learning