检索结果-内蒙古大学图书馆

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

Slax: A Composable JAX Library for Rapid and Flexible Prototyping of Spiking Neural Networks

arXiv

引用

arXiv 2024年

作者： Summe, Thomas M. Joshi, Siddharth Dept. of Computer Science and Engineering University of Notre Dame United States

Recent advances to algorithms for training spiking neural networks (SNNs) often leverage their unique dynamics. While backpropagation through time (BPTT) with surrogate gradients dominate the field, a rich landscape of alternatives can situate algorithms across various points in the performance, bio-plausibility, and complexity landscape. Evaluating and comparing algorithms is currently a cumbersome and error-prone process, requiring them to be repeatedly re-implemented. We introduce Slax, a JAX-based library designed to accelerate SNN algorithm design, compatible with the broader JAX and Flax ecosystem. Slax provides optimized implementations of diverse training algorithms, allowing direct performance comparison. Its toolkit includes methods to visualize and debug algorithms through loss landscapes, gradient similarities, and other metrics of model behavior during training. Copyright © 2024, The Authors. All rights reserved.

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

Developing xApps for Rogue Base Station Detection in SDR-Enabled O-RAN

Developing xApps for Rogue Base Station Detection in SDR-Ena...

引用

2023 IEEE INFOCOM Conference on computer Communications Workshops, INFOCOM WKSHPS 2023

作者： Huang, Jun-Hong Cheng, Shin-Ming Kaliski, Rafael Hung, Cheng-Feng National Taiwan University of Science and Technology Dept. of Computer Science and Information Engineering Taipei Taiwan Research Center for Information Technology Innovation Academia Sinica Taipei Taiwan National Sun Yat-sen University Dept. of Electrical Engineering Kaohsiung Taiwan

ISBN: (纸本)9781665494274

In order to support the diverse requirements of 5G communications, a multitude of RAN components are required. To enable multiple vendor support for 5G, each of whom can independently choose components, Open-RAN (O-RAN) defined a set of standards to which the components must adhere. In addition, O-RAN defines the management elements used to manage each component to secure the 5G networks. While the proposed architecture can manage both 4G and 5G environments, including 5G NSA (Non-Standalone), it inherently suffers from the same vulnerabilities found in 4G LTE. Consequently, an attacker can use unprotected signaling and a low-cost Software Defined Radio (SDR) to launch rogue base station (RBS) attacks on the user equipment (UE), even in O-RAN architectures. In this paper, we consider the stability of signals collected from high-quality operational BSs versus cheap RBSs. Using signal stability features, we develop a machine learning (ML) based RBS detector located on the UE. With the aid of an O-RAN xAPP, ML models can be retrained using the data collected from multiple UEs, and the updated model can be delivered to UEs to enable higher detection accuracy. We conduct extensive experiments by implementing an RBS using a USRP B210, enabling O-RAN using E- Release, and data collected from operational BSs. Moreover, the detector is implemented as an Android APP, which realizes the connection to the O-RAN xAPP. The experimental results show that our detector can achieve more than 99% accuracy, precision 'recall' and F1 score. © 2023 IEEE.

关键词： Base stations

来源：评论

学校读者我要写书评

暂无评论

A comparative study of different scheduling approaches for splittable latency sensitive tasks in Fog-Cloud environment

A comparative study of different scheduling approaches for s...

引用

Ambient Intelligence in Health Care (ICAIHC), International Conference on

作者： Kone Sai Sandeep Challa Aditya Koundinya Adapa Venkata Prabhas Chinmaya Kumar Swain Sambit Kumar Mishra Dept. of Computer Science Engineering SRM University

IoT has revolutionized the way we live and the work we do by connecting different devices through the Internet. In the present scenario, the number of IoT devices are increasing rapidly due to the increase in technology and the increase in the comforts of life. Nowadays we can see that many of them are using IoT devices regularly, it's estimated that by the end of 2030, there will be 30 billion users who will be using IoT applications. These devices send data to the cloud for processing. Due to the distance of the cloud from the IoT devices, the application requests get delayed service responses. So to handle the latency sensitive applications we require the micro cloud service like fog servers deployed near to the data generation points. The fog layer lies between the IoT devices and Cloud which acts as an intermediate layer. This helps in reducing latency of the tasks and provide better performance. As the number of IoT applications keeps on increasing, the resources available with the fog nodes may not handle the upcoming demands. So to overcome these demands, we are using splittable methods to allocate the tasks to Fog/ Cloud nodes more compactly. If a task can be splitted before the deadline into different modules, then we split the given task and allocate those tasks to different fog nodes/ servers and then collecting back the data from the fog nodes/ servers and merging them into a single unit. With the help of this method, we can increase the performance of the system.

关键词：

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

arXiv

引用

arXiv 2024年

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

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/ParallelPerceptionNetwork. © 2024, CC BY.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Image Degradation due to Interacting Hot Pixels and SEUs 37

Image Degradation due to Interacting Hot Pixels and SEUs

引用

37th IEEE International Symposium on Defect and Fault Tolerance in VLSI and Nanotechnology Systems, DFT 2024

作者： Chapman, Glenn H. Farahmandpour, Alireza Chakma, Amit Koren, Israel Koren, Zahava School of Engineering Science Simon Fraser University BurnabyBCV5A 1S6 Canada University of Massachusetts Dept. of Electrical and Computer Engineering AmherstMA01003 United States

ISBN: (纸本)9798350366884

Hot pixels (induced by cosmic rays) in digital imaging sensors accumulate with camera age, impacting the quality of all images produced by the camera. During its lifetime, the imager also experiences Single Event Upsets (SEU s) that create transient defects affecting a single image. The SEUs occur at a much higher rate (events/second) compared to 10-100s additional permanent hot pixels that occur per year. We explore in this paper how an SEU that occurs within a close proximity (within a 5×5 pixel area) of an existing Hot Pixel may distort the image. We also consider situations where an SEU impacts two or more adjacent pixels. When such events happen, the currently employed color demosaicing and JPEG image compression algorithms spread the damage into a 16×16 pixel area, creating significant color changes and resulting in noticeable image degradation. We use formulas developed for a two Hot Pixel interaction to estimate the probability of a Hot Pixel-SEU interaction, and predict its increase with camera age. © 2024 IEEE.

关键词： Cosmic rays

来源：评论

学校读者我要写书评

暂无评论

Breast Cancer Detection with Machine Learning-A Review

Breast Cancer Detection with Machine Learning-A Review

引用

Sustainable Computing and data Communication Systems (ICSCDS), International Conference on

作者： Karnam Meghana Neha Nandal Rohit Tanwar Lipika Goel Gunjan Chhabra Dept. of Computer Science and Engineering Gokaraju Rangaraju Institute of Engineering and Technology Hyderabad India Dept. of Computer Science and Engineering University of Petroleum and Energy Studies Dehradun India Dept. of Computer Science and Engineering Gokaraju Rangaraju Institute of Engineering and Technology Dept. of Computer Science and Engineering Graphic Era Hill University Dehradun India

Malignancy in the breast is a significant public health concern, where timely identification is essential for effective treatment. Machine Learning (ML) and Deep Learning (DL) algorithms are potential tools for prompt detection og breast malignancy through examination of medical images such as mammograms. Convolutional neural networks (CNNs), transfer learning, and ensemble learning are some of the recent techniques being used in this field. Despite the advantages of ML and DL algorithms for breast cancer detection, there are still several challenges that need to be addressed. The lack of diversity in the datasets used to train algorithms is one major challenge, with many datasets based on specific populations that may not represent others. Highly annotated data is also limited in medical field. The objective of this study is to provide researchers with valuable insights and guidance.

关键词： Support vector machines Radio frequency Machine learning algorithms Annotations Transfer learning Sociology Training data

来源：评论

学校读者我要写书评

暂无评论

ML-based Digital Twin for anomaly detection: a case-study on Turtle soccer robots 49

ML-based Digital Twin for anomaly detection: a case-study on...

引用

49th Euromicro Conference on Software engineering and Advanced Applications, SEAA 2023

作者： Liu, Xingyu Van Den Brand, Mark Muctadir, Hossain Muhammad Van De Molengraft, Rene Eindhoven University of Technology Set Cluster Dept. of Mathematics & Computer Science Eindhoven Netherlands Eindhoven University of Technology Department of Mechanical Engineering Eindhoven Netherlands

ISBN: (纸本)9798350342352

In recent years, machine learning (ML) based digital twins (DTs) have seen widespread application in the anomaly detection domain. A search-based literature survey revealed that the majority of the case studies focus on large-scale systems (i.e., nuclear power plant, aerospace, and power grid) producing extensive data. Our work aims to investigate the performance of this technology in smaller-scale systems that generate less data. In this case study, we developed a ML-based DT of the mobility system, the omni wheels, of the Turtle soccer robots. The DT is capable of analyzing historical data collected from the physical robots and differentiating between damaged and undamaged wheels. Our experiments suggest that ML-based DT of small-scale systems is indeed capable of achieving relatively accurate results for anomaly detection use-cases. © 2023 IEEE.

关键词： Sports

来源：评论

学校读者我要写书评

暂无评论

Efficient mmWave Beam Selection using ViTs and GVEC: GPS-based Virtual Environment Capture 14

Efficient mmWave Beam Selection using ViTs and GVEC: GPS-bas...

引用

14th International Conference on Computing Communication and Networking Technologies, ICCCNT 2023

作者： Gupta, Srajan Saxena, Navrati Roy, Abhishek Ravi, Akshay San Jose State University Dept. of Computer Science San Jose United States Mediatek Inc. Engineering San Jose United States

ISBN: (纸本)9798350335095

Millimeter waves (mmWaves) providing higher bandwidth is used by 5G network technology to achieve higher network capacity and faster data transfer. However, the process of beam sweeping across multiple antenna arrays can be time-consuming and inefficient. Machine learning (ML) models can address this issue by predicting the optimal beam pair based on data from on-vehicle light detection and ranging (LIDAR) and global positioning system (GPS) sensors. This paper proposes a new vision transformer (ViT) ML model for beam selection using GPS and LIDAR data. GPS-based virtual environment capture (GVEC) has been introduced to overcome to overcome noise in the LIDAR data caused by adverse weather conditions. The proposed solution demonstrates improved performance compared to previous approaches when tested on noisy LIDAR data, achieving a 92% accuracy in searching among the top 10 beams. © 2023 IEEE.

关键词： Millimeter waves

来源：评论

学校读者我要写书评

暂无评论

Smart Traffic Management System for Urban Mobility Enhancement Using RF Sensors, Cameras, and Machine Learning

Smart Traffic Management System for Urban Mobility Enhanceme...

引用

International Conference on Power, Energy, Control and Transmission Systems (ICPECTS)

作者： Vivekanandan G Pathmanaban J Prabahar Godwin James Barani Sree S Priyanka V Samvardhini V M Dept. of Computer Science and Engineering Sri Sai Ram Institute of Technology

ISBN: (数字)9798331508845

ISBN: (纸本)9798331508852

This paper introduces a Smart Traffic Management System (STMS)employing RF sensors, cameras, and machine learning algorithms to monitor and optimize urban traffic. The system dynamically adjusts traffic signal timings, offers real-time route recommendations based on GPS data, and incorporates adaptive control mechanisms to reduce congestion and improve overall mobility. Simulation studies and real-world testing demonstrate the effectiveness of the STMS in enhancing traffic flow, minimizing waittimes, and contributing to sustainable urbandevelopment.

关键词： Radio frequency Machine learning algorithms Object detection Light emitting diodes Sensor systems Real-time systems Liquid crystal displays Object recognition Intelligent sensors Monitoring

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：