检索结果-内蒙古大学图书馆

7th International Conference on Smart and Sustainable Technologies, SpliTech 2022

作者： Halabi, Talal Abusitta, Adel Carvalho, Glaucio H.S. Fung, Benjamin C.M. University of Winnipeg Dept of Applied Computer Science WinnipegMB Canada School of Information Studies McGill University MontrealQC Canada Brock University Dept of Computer Science and Engineering St. CatharinesON Canada

ISBN: (纸本)9789532901160

With billions of devices already connected to the network's edge, the Internet of Things (IoT) is shaping the future of pervasive computing. Nonetheless, IoT applications still cannot escape the need for the computing resources available at the fog layer. This becomes challenging since the fog nodes are not necessarily secure nor reliable, which widens even further the IoT threat surface. Moreover, the security risk appetite of heterogeneous IoT applications in different domains or deploy-ment contexts should not be assessed similarly. To respond to this challenge, this paper proposes a new approach to optimize the allocation of secure and reliable fog computing resources among IoT applications with varying security risk level. First, the security and reliability levels of fog nodes are quantitatively evaluated, and a security risk assessment methodology is defined for IoT services. Then, an online, incentive-compatible mechanism is designed to allocate secure fog resources to high-risk IoT offloading requests. Compared to the offline Vickrey auction, the proposed mechanism is computationally efficient and yields an acceptable approximation of the social welfare of IoT devices, allowing to attenuate security risk within the edge network. © 2022 University of Split, FESB.

关键词： Fog computing

来源：评论

学校读者我要写书评

暂无评论

Parallel Neural Computing for Scene Understanding from LiDAR Perception in Autonomous Racing

Parallel Neural Computing for Scene Understanding from LiDAR...

引用

Intelligent Systems and Embedded Design (ISED), International Conference on

作者： Suwesh Prasad Sah Suchismita Chinara Dept. Computer Science and Engineering National Institute of Technology Rourkela India

ISBN: (数字)9798331522612

ISBN: (纸本)9798331522629

Autonomous driving in high-speed racing, as opposed to urban environments, presents significant challenges in scene understanding due to rapid changes in the track environment. Traditional sequential network approaches may struggle to meet the real-time knowledge and decision-making demands of an autonomous agent covering large displacements in a short time. This paper proposes a novel baseline architecture for developing sophisticated models capable of true hardware-enabled parallelism, achieving neural processing speeds that mirror the agent's high velocity. The proposed model (Parallel Perception Network (PPN)) consists of two independent neural networks, segmentation and reconstruction networks, running parallelly on separate accelerated hardware. The model takes raw 3D point cloud data from the LiDAR sensor as input and converts it into a 2D Bird's Eye View Map on both devices. Each network independently extracts its input features along space and time dimensions and produces outputs parallelly. The proposed method's model is trained on a system with two NVIDIA T4 GPUs, using a combination of loss functions, including edge preservation, and demonstrates a 2x speedup in model inference time compared to a sequential configuration. Implementation is available at: github/ParallelPerceptionNetwork. Learned parameters of the trained networks are provided at: huggingface/ParallelPercentionNetwork.

关键词： Point cloud compression Solid modeling Laser radar Three-dimensional displays Computational modeling Neural networks Urban areas computer architecture Parallel processing Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Optimized 1D CNNs for Enhanced Early Detection and Accurate Prediction of COPD and Other Pulmonary Diseases

Optimized 1D CNNs for Enhanced Early Detection and Accurate ...

引用

IEEE Region 10 Symposium

作者： Prakash Sahu Sneha Jha Santosh Kumar Computer Science & Engineering IIIT-Naya Raipur Dept. of Data Science CSVTU Chhattisgarh

ISBN: (数字)9798350364866

ISBN: (纸本)9798350364873

Artificial intelligence (AI) advancements have opened new avenues in medical diagnostics, particularly for the early detection of respiratory conditions such as Chronic Obstructive Pulmonary Disease (COPD). COPD, the most common respiratory disease, is marked by breathing difficulties, mucus (sputum) production, coughing, and wheezing. Early diagnosis is crucial for effective management and treatment. A one-dimensional convolutional neural network (1D CNN) model optimized with Adam and RMSprop algorithms is proposed for COPD detection. Using the ICBHI 2017 dataset, the model's performance with each optimizer was evaluated based on training and validation accuracy. The Adam optimizer achieved a training accuracy of 94% and a validation accuracy of 90%. In comparison, the RMSprop optimizer attained a training accuracy of 92% and a validation accuracy of 88%. These findings indicate that the Adam optimizer outperformed RMSprop in terms of accuracy. It showcased the potential of the proposed 1D CNN model as a reliable diagnostic tool, contributing to the advancement of respiratory health analytics.

关键词： Training Analytical models Accuracy Production Chronic obstructive pulmonary disease Convolutional neural networks Reliability Medical diagnosis Artificial intelligence IEEE Regions

来源：评论

学校读者我要写书评

暂无评论

Real-time Human Movement Recognition and Interaction in Virtual Fitness using Image Recognition and Motion Analysis 12

Real-time Human Movement Recognition and Interaction in Virt...

引用

12th International Conference on Awareness science and Technology, iCAST 2023

作者： Yeh, ChunHsiu Shen, Wei-Cheng Ma, Chi-Wei Yeh, Qiu-Tong Kuo, Chung-Wei Chen, Jong-Shin Dept. of Information Engineering Computer Science Feng Chia University Taichung Taiwan CollegeofComputerScience Jimei University Xiamen China Dept. of Information Communication Engineering ChaoYang University of Technology Taichung Taiwan

ISBN: (纸本)9798350324693

This research focuses on the application of image recognition, motion detection, and artificial intelligence techniques to achieve real-time identification and analysis of human movements in a dynamic manner. The primary goal is to establish a seamless association between captured human movements and corresponding actions performed by virtual avatars. The implementation involves the utilization of image processing methodologies from python's OpenCV and pygame libraries. MediaPipe technology is employed to capture users' images and facilitate accurate detection of specific movement patterns. By calculating the angles between users' limbs and torso, the correctness of certain movements can be assessed. The research is practically developed and applied to a simulated fitness system, where an interactive exercise game platform is devised. The system is characterized by its engaging user interface, enhancing the ease and enjoyment of fitness activities. The seamless integration of interactive virtual avatars and real users within physical activities serves to stimulate greater participation, particularly among the elderly. The amalgamation of image recognition and motion detection technologies presents opportunities for combining fitness and entertainment in diverse life scenarios, offering an enhanced experience of physical exercise and enjoyment. © 2023 IEEE.

关键词： Image recognition

来源：评论

学校读者我要写书评

暂无评论

Gaussianity and the Kalman Filter: A Simple Yet Complicated Relationship

arXiv

引用

arXiv 2024年

作者： Uhlmann, Jeffrey K. Julier, Simon J. Dept. of Electrical Engineering and Computer Science University of Missouri-Columbia United States Department of Computer Science University College London United Kingdom

One of the most common misconceptions made about the Kalman filter when applied to linear systems is that it requires an assumption that all error and noise processes are Gaussian. This misconception has frequently led to the Kalman filter being dismissed in favor of complicated and/or purely heuristic approaches that are supposedly "more general" in that they can be applied to problems involving non-Gaussian noise. The fact is that the Kalman filter provides rigorous and optimal performance guarantees that do not rely on any distribution assumptions beyond mean and error covariance information. These guarantees even apply to use of the Kalman update formula when applied with nonlinear models, as long as its other required assumptions are satisfied. Here we discuss misconceptions about its generality that are often found and reinforced in the literature, especially outside the traditional fields of estimation and control. Copyright © 2024, The Authors. All rights reserved.

关键词： Kalman filters

来源：评论

学校读者我要写书评

暂无评论

Perturb, Attend, Detect and Localize (PADL): Robust Proactive Image Defense

arXiv

引用

arXiv 2024年

作者： Bartolucci, Filippo Masi, Iacopo Lisanti, Giuseppe Computer Science and Engineering Dept. CVLab University of Bologna Bologna Italy Computer Science Dept. OmnAI Lab Sapienza University of Rome Rome Italy

Image manipulation detection and localization have received considerable attention from the research community given the blooming of Generative Models (GMs). Detection methods that follow a passive approach may overfit to specific GMs, limiting their application in real-world scenarios, due to the growing diversity of generative models. Recently, approaches based on a proactive framework have shown the possibility of dealing with this limitation. However, these methods suffer from two main limitations, which raises concerns about potential vulnerabilities: i) the manipulation detector is not robust to noise and hence can be easily fooled;ii) the fact that they rely on fixed perturbations for image protection offers a predictable exploit for malicious attackers, enabling them to reverse-engineer and evade detection. To overcome this issue we propose PADL, a new solution able to generate image-specific perturbations using a symmetric scheme of encoding and decoding based on cross-attention, which drastically reduces the possibility of reverse engineering, even when evaluated with adaptive attacks [31]. Additionally, PADL is able to pinpoint manipulated areas, facilitating the identification of specific regions that have undergone alterations, and has more generalization power than prior art on held-out generative models. Indeed, although being trained only on an attribute manipulation GAN model [15], our method generalizes to a range of unseen models with diverse architectural designs, such as StarGANv2, BlendGAN, DiffAE, StableDiffusion and StableDiffusionXL. Additionally, we introduce a novel evaluation protocol, which offers a fair evaluation of localisation performance in function of detection accuracy and better captures real-world scenarios. Copyright © 2024, The Authors. All rights reserved.

关键词： Reverse engineering

来源：评论

学校读者我要写书评

暂无评论

Deep Learning for Tea Leaf Disease Detection: Comparing VGG16, Vision Transformer, EfficientNetB3, and Xception

Deep Learning for Tea Leaf Disease Detection: Comparing VGG1...

引用

International Conference on Computing and Networking Technology (ICCNT)

作者： Rokonozzaman Ayon Arif Shahriar Mubin Al-Amin Gazi Hridoy Mayen Uddin Mojumdar Narayan Ranjan Chakraborty Md. Jahidul Islam Dept. of Computer Science & Engineering Daffodil International University Dhaka Bangladesh

ISBN: (数字)9798350370249

ISBN: (纸本)9798350370270

The need for efficient disease management in crop production is highlighted by the economic relevance of agriculture. Various leaf diseases cause major losses for tea, a staple product in many places. In this paper, a deep learning-based method is provided for the automated identification of illnesses affecting tea leaves. Using the unique architectural benefits of four models: VGG16, ViT (Vision Transformer), EfficientNetB3, and Xception, their performance is examined. Each model’s accuracy and effectiveness in classifying diseases are evaluated by conducting a thorough series of experiments on a dataset of photos of tea leaves. The results show that ViT and VGG16 perform better than expected, with accuracies of $\mathbf{9 6. 0 9 \%}$ and $\mathbf{9 8. 2 9 \%}$, respectively. These results highlight the effectiveness of transformer-based designs and conventional CNN in the identification of tea leaf disease. Our research also demonstrates how deep learning approaches can improve agricultural practices by enabling early and accurate disease identification, which would facilitate proactive disease control tactics and increase crop output.

关键词： Deep learning Learning systems computer vision Accuracy Biological system modeling Crops Production Transformers Reliability Diseases

来源：评论

学校读者我要写书评

暂无评论

Addressing Data Imbalance in Machine Learning Solutions for Sexual Violence Detection

Addressing Data Imbalance in Machine Learning Solutions for ...

引用

International Seminar on Research of Information Technology and Intelligent Systems (ISRITI)

作者： Azani Cempaka Sari Karto Iskandar Amrita Prasad Risma Yulistiani Aditya Kurniawan Nora Fitriawati Setiawan Joddy Rico Halim Computer Science Department School of Computer Science Bina Nusantara University Jakarta Indonesia Dept. of Mechanical Engineering Jadavpur University Kolkata India

ISBN: (数字)9798331519643

ISBN: (纸本)9798331519650

The case of sexual violence has become rampant especially with the introduction of the internet, social. Platforms act both as a means of perpetuating this violence through harassment of victims and as a platform for fighting back by sharing incidences of such occurrences. It is immediately crucial to obtain highly effective and accurate classification models for early identification and control of this behavior. This research aims to address data imbalance in detecting text-based sexual harassment in Indonesia using machine learning techniques by finding better ways of mitigating the imbalance that exists and where non-harassment cases dominate the dataset rather than harassment cases. This study is based on data extracted from Instagram, TikTok, and X (previously Twitter) with the Synthetic Minority Over-sampling Technique (SMOTE) to overcome the imbalance of data. Specifically, three machine learning algorithms were used: Support Vector Classifier (SVC), Random Forest Classifier, and Naïve Bayes Classifier were used. The SVC model had the highest accuracy of 97 percent under the accuracy, precision, recall, and F1-score categories. The findings clearly suggest that when applied with SMOTE as a preprocessing and SVC classifier, the problem of data imbalance is solved and harassment is detected more accurately. This paper further advances the area of Artificial Intelligence in online harassment, providing a novel approach to reach safer internet space and improve the assistance to the victims.

关键词： Accuracy Machine learning algorithms Video on demand Social networking (online) Static VAr compensators Vectors Safety Bayes methods Web sites Random forests

来源：评论

学校读者我要写书评

暂无评论

Enhancing Agricultural Quality Control: Freshness Detection in Root Vegetables Through Deep Neural Networks

Enhancing Agricultural Quality Control: Freshness Detection ...

引用

International Conference on Computing and Networking Technology (ICCNT)

作者： Aklema Akther Shorna Abida Sultana Ave Hasibul Hasan Chowdhury Mahi Sabrina Bari Shital Nourin Jahan Sowrna Aynul Hasan Nahid dept. computer science and engineering Daffodil International University Dhaka Bangladesh

ISBN: (数字)9798350370249

ISBN: (纸本)9798350370270

In monarchy agricultural technology, this research pioneers the classification of freshness in root vegetables, a field that has been abruptly neglected in earlier research on a variety of vegetables. This study addresses this research gap and employs a transfer learning-based system to swiftly determine the freshness of root vegetables through images. To conduct this research, we create our own dataset which contains six different vegetables, “Potato”, “Radish”, “Carrot”, “Onion”, “Taro”, and “Turnip”. Vegetables were collected from the local market and clicked manually by personal mobile phone, giving the dataset authenticity and real-world application. In total we collect 6384 images, divided into 3 classes: “Fresh”, “Aged” and “Rotten” based on their physical attributes and texture. Different types of transfer learning model VGG19, ResNet50, InceptionResNetV2 and DenseNet201 used to classify freshness of root vegetables. Although every model achieved decent accuracy, DenseNet201 achieved a remarkable result. DenseNet201achived highest accuracy 98.75% where data loss was 0.02%. Conversely, though ResNet50 achieved lowest accuracy 60.97% among all models. The variety of algorithms used seeks to determine the best course of action for this particular agricultural setting in addition to precisely classifying freshness. This research not only defines a novel benchmark but also sets the stage for future explorations in agricultural technology, ushering in a new era of precision farming for root vegetables

关键词： Deep learning Accuracy Image processing Transfer learning Quality control Aging Mobile handsets Classification algorithms Residual neural networks Farming

来源：评论

学校读者我要写书评

暂无评论

Seamless Service Evolution: Enhancing Customer Satisfaction Using AWS-Driven AI Chatbots for Restaurant Ordering

Seamless Service Evolution: Enhancing Customer Satisfaction ...

引用

Sustainable Expert Systems (ICSES), International Conference on

作者： Mohammed Jaffer Ali M R V Vyshnavi Konda V S Harshith Kumar Sravya Vishnubhatla Shinu M Rajagopal Dept. of Computer Science and Engineering Amrita School of Computing Bengaluru India

ISBN: (数字)9798331540364

ISBN: (纸本)9798331540371

In the current restaurant business environment, factors have emerged in the manner in which consumers are attended to and the ways orders are placed. As good as these traditional systems may be, they do not offer the optimum flexibility and consumer convergence that today's client demands for fast and convenient services. Recent approaches using only basic form of automation and rule-based chatbots do offer some enhancements but lack adequate capabilities to deal with huge volumes of orders, custom recommendations and scalability according to demand. Further, the existing systems have problems with scalability, versatility to different customers' trends and needs, and real-time interactions. To overcome these challenges our work, will utilize modern advances in artificial intelligence and cloud technology. The proposed system uses Amazon Lex for NLU, A WS Lambda for processing multiple booking requests simultaneously without any intervention from the users and Amazon Polly for making interaction more interactive with personalized audio. In addition, Amazon Rekognition and A WS S3 for images will be used for analysis of images and proper storage of data respectively, enhancing the customer satisfaction by recommending products based on a past order. All these technologies integrated with A WS ecosystem in our solution will provide convenience, efficiency and passion to restaurant business proprietors to fulfill the increasing demands of modern customer. The system also supports linkage with some commonly used applications such as WhatsApp, guaranteeing that the customers receive timely updates. Thus, this solution is designed to create an improved status of restaurant ordering and eliminate the problematic experiences placed in current schemes.

关键词： Scalability Customer satisfaction Chatbots Market research Real-time systems Image storage Artificial intelligence Internet telephony Optimization Business

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：